Lightweight Construction and
Composite Technology
Master Thesis
Evaluation and Development of a
Continuous Longitudinal Joining Process for
Carbon-Reinforced High-Performance
Composites in Aircraft Primary Structure
Faculty of
Mechanical and
Process
Engineering
Alexander Sänger, B.Eng.
Alexander Sänger, B.Eng.
Schornstraße 11
D-81669 München
+49 89 447 03 78
alexander.saenger1@hs-augsburg.de
Matricualtion Number: 95 27 04
First Examiner: Prof. Dr.-Ing. André Baeten
Second Examiner: Prof. Dr.-Ing. Ralf Goller
Supervisor: Dr. Stefan Jarka
Topic received on: 22/05/2017
Date of submission: 13/11/2017
A disclosure agreement is applied
The copyright belongs to the author of this work.
Hochschule für
angewandte Wissenschaften
University of Applied Sciences
An der Hochschule 1
D-86161 Augsburg
Telefon +49 821 55 86-0
Fax +49 821 55 86-3222
www.hs-augsburg.de
info@hs-ausgsburg.de
To
my parents
& Urli
Hochschule Augsburg
University of Applied Sciences Augsburg
M A S T E R T H E S I S
Evaluation and Development of a Continuous Longitu-
dinal Joining Process for Carbon-Reinforced High-Per-
formance Composites in Aircraft Primary Structure
Alexander Sänger, B.Eng.
Discussed: // / Received: // / Submission: //
©The Author . This work is restricted by a disclosure agreement.
Abstract Although (carbon) compos-
ites seem to be arrived in aircraft struc-
tures, the full lightweight potential is
not reached and state of the art pro-
cesses are usually not bre-fair. In ad-
dition, most frequently used are ther-
moset matrices which require energy
demanding material cooling and cur-
ing, high added-value processing ma-
terials and labour-intensive process
steps. This work presents the state
of the art aircraft manufacturing with
spotlights on the main challenges and
potentials for thermoplastic resin sys-
tems. An overview and evaluation of
fusion bonding approaches is given
leading to ultrasonic welding as pro-
cess of choice. Theoretical consid-
erations on heating mechanisms lay
the foundation for subsequent para-
metric study of statically ultrasonically
welded CF/PEEK specimen. Observa-
tions and results are discussed and
transferred to an continuous welding
concept for robotic application. There-
fore, this work does not only provide a
technological overview and experimen-
tal data on so far rather neglected
ultrasonic welding of CF/PEEK com-
posites, but also suggestions and rst
steps towards the aircraft manufactur-
ing of the future.
Keywords Ultrasonic Welding Ther-
moplastic Composites Heating Mech-
anisms Continuous Joining Robot
Application Aircraft Manufacturing
Alexander Sänger
alexander.saenger@hs-augsburg.de
Faculty of Mechanical Engineering
University of Applied Sciences
D-866 Augsburg, Germany
Restriction Note
The presented master thesis contains condential information. All remarks, data and
results of the presented thesis must be treated as condential and must not be available
to third parties without written consent of Deutsches Zentrum für Luft- und Raumfahrt
e.V. (DLR) and the author. Disclosure, publishing or duplication of the thesis in parts or
in extracts even in digital form are not permitted without expressly authorisation.
This thesis is liable to a prohibition of publishing, restriction in access and searching
for persons who are not in charge of evaluating this thesis. Both parties, author and
supervisors, agreed upon a conspiracy of silence.
Alexander Sänger, B.Eng.
Schornstraße 
866 München
Dr.-Ing. Stefan Jarka
Deutsches Zentrum für
Luft- und Raumfahrt e.V.
Am Technologiezentrum
86 Augsburg
Alexander Sänger, B.Eng.
Schornstraße 
866 München
Prof. Dr.-Ing. André Baeten
Hochschule für angewandte
Wissenschaften Augsburg
An der Hochschule
866 Augsburg
II
Table of Content
Page
Restriction Note II
List of Figures VII
List of Tables X
List of Abbreviation XI
List of Symbols XIII
Introduction
. Market Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. Clean Sky Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. Main Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. Thesis’ Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
State of the Art, Potentials and Criteria
. Aircraft Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. Composites Deployment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Thermosets (TS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Thermoplatics (TP) . . . . . . . . . . . . . . . . . . . . . . . . . . 8
. Main Drawbacks and Challenges . . . . . . . . . . . . . . . . . . . . . . . 
.. Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Joining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Recycling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Evaluation Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Prerequisites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Process Capability . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Aircraft Applicability . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Other Secondary Criteria . . . . . . . . . . . . . . . . . . . . . . . 
Overview on Fusion Joining Processes 
. Bulk Heating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Two-stage Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
. Frictional Heating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Various Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
III
Table of Content
.. Ultrasonic Welding . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Electro-Magnetic Heating . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Various Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Resistance Welding . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Induction Welding . . . . . . . . . . . . . . . . . . . . . . . . . . . 
Evaluation 
. Process Capability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Process Chain Adoption. . . . . . . . . . . . . . . . . . . . . . . . 
. Aircraft Applicability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Certication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
. Secondary Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Fibre-Fairness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Heating Characteristics . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
.. Environmental Aspects . . . . . . . . . . . . . . . . . . . . . . . . 6
. Resumée and Final Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Pre-Testing Campaign 68
. First Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
. Second Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
. Third Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
6 Heating Models, Mechanisms and Parameters 
6. Enthalpy of Fusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
6. Deformation Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
6. Interfacial Friction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
6. Intermolecular Friction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
6. Combined Heating Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . 6
6.6 Multi-Body Dynamics and Interfacial Friction . . . . . . . . . . . . . . . . . 8
6. Composite Heat Flow Behaviour . . . . . . . . . . . . . . . . . . . . . . . 
6.8 Theory Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
6. Scientic Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Experimental Set-Up 8
IV
Table of Content
. Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Laminate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Energy Directors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Test Specimen Design . . . . . . . . . . . . . . . . . . . . . . . . . 8
. Manufacturing Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Ultrasonic Welding Machine . . . . . . . . . . . . . . . . . . . . . 8
.. Sonotrode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Fixture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
. Analysis Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Manual Bending/Breaking . . . . . . . . . . . . . . . . . . . . . . . 8
.. Lap Shear Tension Test . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Fracture Surface Analysis . . . . . . . . . . . . . . . . . . . . . . . 8
8 Parametric Study, Results and Discussion 88
8. Anvil Stiffness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8. Specimen Arching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
8. Edge Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8.. Unbalanced Clamping Lateral . . . . . . . . . . . . . . . . . . . 
8.. Unbalanced Clamping Longitudinal . . . . . . . . . . . . . . . . 
8.. Edge Concentration Conditions . . . . . . . . . . . . . . . . . . . . 
8. Patch Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8.. Guided Melt Initiation . . . . . . . . . . . . . . . . . . . . . . . . . 
8.. Interfacial Friction . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8. Failure Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8.6 Heat Flow Behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8.6. In-Plane Direction . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8.6. Thickness Direction . . . . . . . . . . . . . . . . . . . . . . . . . . 6
8. Parametric Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8.. Obtained Experimental Data . . . . . . . . . . . . . . . . . . . . . 
8.. Force-Elongation-Curves . . . . . . . . . . . . . . . . . . . . . . . 
8.. Vibration Time-Force-Correlation . . . . . . . . . . . . . . . . . . . 
8.. Vibration Time-Amplitude-Correlation . . . . . . . . . . . . . . . . 
8.. Weld Area-Collapse-Correlation . . . . . . . . . . . . . . . . . . . . 
8..6 LSS-Energy Density-Correlation . . . . . . . . . . . . . . . . . . . . 
8.. LSS-Mean Power-Correlation . . . . . . . . . . . . . . . . . . . . . 
8..8 LSS-Vibration Time-Correlation . . . . . . . . . . . . . . . . . . . . 
8.. Comparative Lap Shear Strength / Weld Factor . . . . . . . . . . . 
8.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
V
Table of Content
Continuous Welding Concept Development 6
. Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
. Anvil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
. Robot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Endeffector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
.. Process Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Equipment Weight Impact . . . . . . . . . . . . . . . . . . . . . . . 
.. Mass Moment of Inertia . . . . . . . . . . . . . . . . . . . . . . . . 
. Ultrasonic Welding Equipment . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Generator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.. Sonotrode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.6 Energy Directors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
 Conclusion 
. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Acknowledgements 8
Appendices 
A Experimental Set-Up Addendum 
A. Computed Maximum Overlap Length . . . . . . . . . . . . . . . . . . . . . 
A. Laminate/Film Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 
B Statistical Evaluation 
C Fracture Surface Microscopic Analysis 
D Clamping Force Estimation 6
About the Author 8
Bibliography 
Declaration of Authorship 
VI
List of Figures
. Composite Content in Aviation Structures 6- . . . . . . . . . . . . .
. Primary Aircraft Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. Share of Structural Materials of Airbus Aircraft . . . . . . . . . . . . . . . . . 6
. Schematic Young’s Modulus over Temperatures for the Different Plastic
Classications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. Glass Transition/Melting Temperature of Different PEK
n
Derivatives . . . . .
. MG Bracket Rear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.6 Aircraft Repair Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Current Automated Fibre Placement Production Method . . . . . . . . . . . 8
.8 New Developments in A Structural Design . . . . . . . . . . . . . . . . . 
. Primary and Secondary Evaluation Criteria . . . . . . . . . . . . . . . . . . . 
. Overview and Classication of Joining Techniques . . . . . . . . . . . . . . 
. Process Window of Thermabond
®
Process . . . . . . . . . . . . . . . . . . . 6
. Two-Stage Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Ultrasonic Welding Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . 
. Specic Heat over Temperature for Thermoplastic Types . . . . . . . . . . . 
.6 Variants of Ultrasonic Welding . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Ultrasonic Seam Welder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
.8 Ultrasonic Fibre Placement Head and its Application . . . . . . . . . . . . . 6
. Schematic Continuous Microwave Joining Set-Up . . . . . . . . . . . . . . . 
. Schematic Dielectric Joining Set-Up . . . . . . . . . . . . . . . . . . . . . . . 8
. Working Principle of Resistance Welding . . . . . . . . . . . . . . . . . . . . 
. Resistance Welding Set-up and Process Steps . . . . . . . . . . . . . . . . . 
. Continuous Resistance Welding . . . . . . . . . . . . . . . . . . . . . . . . . 
. Various Coil Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Comparative LSS of Ultrasonic, Resistive and Inductive Welded CF/PPS
Specimen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
. Averaged LSS values for APC- (CF/PEEK) Specimen . . . . . . . . . . . . . 6
. S-N Curves of Differently Welded CF/PPS Specimen . . . . . . . . . . . . . . 
. Failure Modes in Lap Shear Tests . . . . . . . . . . . . . . . . . . . . . . . . 8
. Magnetic Flux Concentrator . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
VII
List of Figures
.6 ED Temperature Development during Ultrasonic Welding for Different Con-
gutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
. Typical Temperature-Time-Curve of the Continuous Induction Welding . . . 6
.8 Temperature Development over Time for Different Welding Speeds . . . . . 6
. First Pre-Testing Campaign Set-Up . . . . . . . . . . . . . . . . . . . . . . . . 68
. Third Pre-Test Overlap Conguration . . . . . . . . . . . . . . . . . . . . . . 
. Spot-Welded Overlap Specimen . . . . . . . . . . . . . . . . . . . . . . . . . 
6. Energy Losses in Ultrasonic Welding Process (Dukane , ) . . . . . . . 
6. Storage (E’) and Loss (E”) Moduli for PEEK at  kHz . . . . . . . . . . . . . 6
6. Melting Time of PMMA over Contact Pressure for Different Amplitudes . . . 
6. Investigated ED Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
6. Composite Wall Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
. Test Specimen Dimensions acc. to ASTM D  . . . . . . . . . . . . . . . 8
. Fixture/Anvil Variants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
. Test Stand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
. Ultrasonic Prexation of Two Loose PEEK lms on Laminate with Handheld
Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
. Steel Table Anvil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
.6 Manual -Point-Bending Test . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
8. ANSYS FEM Harmonic Response Simulation for Total Deformation of De-
ployed Anvil Variations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8. Combined Aluminium-Steel Solid Fixture Design . . . . . . . . . . . . . . . . 8
8. Clamping-Tilting Issue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
8. Initial Line Contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8. Melt Flow from Inner Edge due to Initial Line Contact . . . . . . . . . . . . . 
8.6 Sideward Edge Effect due to Longitudinal Unbalanced Clamping . . . . . . . 
8. Combined Lateral/Longitudinal Unbalanced Clamping Edge Effect . . . . . 
8.8 Preferential Heating due to Edge Unregularities . . . . . . . . . . . . . . . . 
8. Lifting Effect of ED Patches for Avoiding Edge Effects . . . . . . . . . . . . . 
8. Dry Fibres Indicate Lack of Matrix . . . . . . . . . . . . . . . . . . . . . . . . 
8. Fracture Surfaces with Indication of Guided Melt Initiation . . . . . . . . . . 
8. Interlayer Melting Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8. Observed Failure Modes on Fracture Surfaces . . . . . . . . . . . . . . . . . 
8. Observed Failure Modes on Fracture Surfaces . . . . . . . . . . . . . . . . . 6
VIII
List of Figures
8. Microscopical Analysis of Melt Front Propagation In/Transverse Fibre Di-
rection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
8.6 Force over Traverse Path (with Specimen ID) . . . . . . . . . . . . . . . . . . 8
8. Fracture Surfaces with Marked Joint Area . . . . . . . . . . . . . . . . . . . 8
8.8 Peel Stress Distribution (FAA ) . . . . . . . . . . . . . . . . . . . . . . . 
8. LSS over Welded Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8. Vibration Time over Weld Force . . . . . . . . . . . . . . . . . . . . . . . . . 
8. Vibration Time over Weld Force . . . . . . . . . . . . . . . . . . . . . . . . . 
8. Vibration Time over Weld Force . . . . . . . . . . . . . . . . . . . . . . . . . 
8. LSS over Energy Density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8. LSS over Mean Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8. LSS over Vibration Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
8.6 Torn Interlaminar Lap Shear Specimen . . . . . . . . . . . . . . . . . . . . . 
. Aircraft Shell Joining with Ultrasonic “Black Box” Endeffector . . . . . . . . 6
. Payload Diagram for KR - PA . . . . . . . . . . . . . . . . . . . . . . . . 
. Existing Patents on Endeffector Concepts . . . . . . . . . . . . . . . . . . . 8
. Endeffector Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Minimum Working Example Draft . . . . . . . . . . . . . . . . . . . . . . . . 
.6 Triangular Energy Director Investigation . . . . . . . . . . . . . . . . . . . . . 
. Quality Management Schematic . . . . . . . . . . . . . . . . . . . . . . . . . 
D. Required Force Dependency on Roller Radius and Width with Virtual  N
Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
Note: All pictures, gures and graphics, as far as not indicated in captions, are taken or
created by the author and the copyright belongs to him.
IX
List of Tables
. Mechanical Properties of Selected Semi-Crystalline TP and TS . . . . . . . .
. Comparative Invar6 and Ceramic Properties . . . . . . . . . . . . . . . . . . 
. Mechanical and Thermal Properties of PPS and PEEK . . . . . . . . . . . . . 
. Process Parameters for Ultrasonic, Induction and Resistance Welding . . . 
. Interlaminar Fracture Toughness G
Ic
for Different Joining/Material Cong-
urations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Evaluation Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
6. Parameter Inuence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
8. Weld Strengths and Factors for Material Combinations . . . . . . . . . . . . 
8. Obtained Experimental Data . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Equipment Weight Impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
. Technical Data of Available Ultrasonic Generators . . . . . . . . . . . . . . . 
A. Hauer CF/PEEK Plate Properties . . . . . . . . . . . . . . . . . . . . . . . . 
A. TORAYCA
®
T Carbon Fibre Properties . . . . . . . . . . . . . . . . . . . . 
A. Aptiv
®
 black PEEK Film Properties . . . . . . . . . . . . . . . . . . . . . 
A. LITE
®
TK PEEK Film Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 
B. Student’s t-Distribution Values . . . . . . . . . . . . . . . . . . . . . . . . . . 
X
List of Abbreviation
AFP Automated Fibre Placement
ALM Additive Layered Manufacturing
APC Aromatic Polymer Composite
APC- PEEK matrix with AS carbon bre reinforcement
ARW Automated Resistance Welder
AS HexTowcontinuous, high strength/strain, PAN based carbon bre
ASTM American Society for Testing Materials
BVID Barely Visible Impact Damage
CF Carbon Fibre
CFRP Carbon Fibre Reinforced Plastics
CG Centre of Gravity
CMC Ceramic Matrix Composites
CRW Continuous Resistance Welding
CTE Coefcient of Thermal Expansion
DCB Double Cantilever Beam
DIN Deutsches Institut für Normung (German Institute for Standardisation)
DSG Design Service Goal
EASA European Aviation Safety Agency
EBM Electron Beam Melting
ED Energy Director
EN European Normative
FAA Federal Aviation Administration
FSM Friction Stir Welding
HAZ Heat-Affected Zone
HDT Heat Deection Test
IM Intermediate Modulus
IM HexTowcontinuous, high performance, intermediate modulus, PAN based
carbon bre
IR Infrared
IRW Infrared Welding
. Mattia 
XI
List of Abbreviation
IW Induction Welding
LBW Laser Beam Welding
LSS Lap Shear Strength
LTW Laser Transmission Welding
MMC Metal Matrix Composites
MWE Minimum Working Example
NDI Non-Destructive Inspection
NDT Non-Destructive Testing
PA Polyamide
PAN Polyacrylonitrile
PAX Passenger
PEI Polyetherimide
PEEK Polyetheretherketone
PTFE Polyetetraourethylene (aka Teon
®
)
PU Polyurethane
PVC Polyvinylchloride
RC Recurring Costs
RF Radio Frequency
RW Resistance Welding
SLM Selective Laser Melting
SRW Sequential Resistance Welding
TP Thermoplastic
TPC Thermoplastic Composite
TS Thermoset
UD Unidirectional
ul Unit Length
US Ultrasonic (here also: Ultrasonic Welding)
XII
List of Symbols
in basis SI unit system
α
K
Resistance-Temperature Coefcient
α
h
Hammering Coefcient
a
S,0
m Sonotrodes Peak-to-Peak Amplitude
A m
2
Area
c
J
kg·K
Specic Heat / Thermal Capacity
D
q
N·kg
m
Damping Coefcient
η Pa ·s Dynamic Viscosity
η kg ·m Damping Coefcient
η
cpl
Coupling Efciency
ε Elongation
ˆ
ε Amplitude Strain
E, E
0
N
m
2
, Pa Young’s Modulus (Elastic/Storage Modulus)
E
00
N
m
2
, Pa Loss Modulus
E
N
m
2
, Pa Complex Elastic Modulus
f Hz Frequency
F
C
N Contact Force
F
N
N Normal Force
F
τ
N Interfacial Friction
γ Shear Strain
g
m
s
2
Gravitational Constant (9.8065
m
/s
2
)
γ Specic Electrical Resistance
G
Ic
J
mm
2
Interlaminar Fracture Toughness Mode I
H(I)
A
m
Current Dependent Magnetic Field
h
m
J
m
2
Specic Fusion Enthalpy
H
m
J Fusion Enthalpy
I A Electrical Current
k
z
W
m
2
·K
Thermal Resistance Coefcient
XIII
List of Symbols
λ
W
m·K
Thermal Conductivity
λ
i j
W
m·K
Thermal Conductivity Tensor (nd Order)
L, l m Length
µ
H
m
Magnetic Permeability
µ Constant Sliding Coefcient
m kg Mass
M
A
Nm Torque
Gradient Operator
ν Poissons Ratio
ω
rad
s
=
q
N
m·kg
Angular Frequency
ϕ Fibre Volume Content
p
N
m
2
, Pa Pressure
P W (Electrical) Power
˙q
W
m
2
Heat Flow/Flux
˙
Q
f ric
W Friction Dissipated Power
˙
Q
in
W Input Heat Power
˙
Q
mc
W Intermolecular Dissipated Power
ρ
m
Electrical Conductivity (Specic Electrical Resistance)
ρ
kg
m
3
Density
ρ
1
, ρ
2
m Curvature Radius
R Electrical Resistance
R
a
m Surface Roughness (Arithmetical Mean Deviation)
R
m
N
m
2
, Pa Ultimate Tension Strength
σ
N
m
2
, Pa Stress
σ
N
N
m
2
, Pa Normal Stress
σ
yy
N
m
2
, Pa Vertical Stress on Horizontal Interface
τ
N
m
2
, Pa Shear Stress
T °C/°K Temperature
T
d
°C Decomposition Temperature
T
g
°C/°K Glass Transition Temperature
T
m
°C/°K Melting Temperature
XIV
List of Symbols
T
s
°C/°K Solidus Temperature (=T
m
)
T
SC
°C/°K Crystalline Melting Temperature
δu
m Virtual Displacement
U V Electrical Voltage
v
m
s
Velocity
w m Width
W
d
J Deformation Work
Indices
k In Fibre Direction
Perpendicular/Transverse Fibre Direction
f Fibre
m Matrix
Coordinate Systems
x, z Global Robot Coordinate System
ξ , ζ Local Coordinate System
XV
Introduction
. Market Review
In the last decades, aviation industry expedited a change climaxing in the recent years
with the newest aircraft types of the market leaders A and B. The A and
8 Dreamliner, respectively, are the rst aircraft in history manufactured predominantly
of composite materials. Both are the outcome of trend which has been growing for
about  years (Figure .).
1965 1975 1985 1995 2005 2015
15
30
45
60
F-
F-
F-
F-6
F-8C/D
B-
F-
Euroghter
AM
A
A
A-
A-6
A8
A
B
B6
B-
B-
B
B8
MD8
MD
Advanced Composite Content [%]
Fighter/Military
Commercial Airbus
Commercial Boeing
Commercial MD
Figure . Composite Content in Aviation Structures 6- (Data: Teßmer 6, ; Noor
, )
Initially discovered due to its lightweight and strength potentials, the utilisation of
composite materials is these days driven by the pursuit of developing more efcient
products concerning fuel consumption and air pollution, incorporating improvements in
payload capacity and operating range. Such solutions are received with open arms by
pressured airlines competing with low-cost carriers on the market.
Originally used only for secondary structures
, composites have nowadays become
the backbone of a modern aircraft forming essential fuselage and wing components. Al-
though lightweight potentials are by far not exhausted due to certication requirements,
aircraft performance has improved impressively with the development of various resin
. non-load bearing structures that do not cause immediate danger upon failure
: Introduction
and bre types. Starting with the invention of epoxy resins in the s, followed by
the invention of glass bres in 6 and carbon bres in , the corner stone for
high strength applications was laid (Gerdeen et al. , ). Today, it is not possible
to imagine aircraft manufacturing without carbon, glass or aramid bres often times in
combination with polymeric matrix.
. Clean Sky Project
Apart from strongly risen number of passengers over the last years (UN DESA and Boe-
ing ) and the predicted ongoing increase in air trafc and eet size (Boeing 6;
DLR , ), there is a growing awareness of climate change, too, contrasting these
trends. Focal points lie on the need to reduce greenhouse gases and further sources of
environmental pollution without hampering efforts of a globalised market and society
as this achievement is essentially based on developments and affordability of ights all
over the globe. Furthermore it is not only about contributing to environmental sustain-
ability but also to support European economys competitiveness.
Picking up this trend, as part of the EU Horizon  research and innovation pro-
gramme by the European Union Funding for Research & Innovation, Clean Sky was founded
as an initiative of the European Commission and the European aviation industry aimed
at developing innovative, cutting-edge technology for future aircraft with reduced CO
,
gas emissions and noise levels. The new Clean Sky Programme is scheduled from
 to , in total,  institutions from  countries participate in this joint under-
taking with a total budget of e billion (Clean Sky JU 6).
This work is located within the work stream conducted by A and the G
A C (DLR) targeted at developing a complete new approach of aircraft
with thermoplastic primary structure and associated processes.
The DLR Institute of Structures and Design investigates in Augsburg and Stuttgart
developments of thermoplastic high-performance structures focussing on holistic ob-
servation of a continuous process chain. This ranges from development of the con-
struction and production technology to exible automation solutions for production of
carbon bre reinforced thermoplastics.
2
: Introduction
Especially fusion bonding processes offer an opportunity for thermoplastic structural
parts since they provide high-strength, areal and dense joints with a concurrent high po-
tential for automation. So far, these joining methods have not been deployed in aircraft
series production yet. In particular for the longitudinal seams of aircraft skins, special
challenges arise.
. Main Challenges
Although this trend of composites comes along with advantageous properties es-
pecially low density and thus high specic strength –, there are disadvantages which
must not be neglected. Aviation industry is facing new challenges with this promising
material type.
Many manufacturing steps for composites are still completely manual, demand skilled
workers and much high quality process materials which are wasted after curing. All in
all, it is still a time, material and cost intensive manufacturing compared to conservative
metal constructions.
Especially the joining approach is borrowed from well-known process in metal air-
craft production, e.g. A family. Mostly deployed joining type even for composite
structures is conventional riveting
which is not adequate for brous compounds.
As consequence, maintenance and repairability become complicated and costly if
even possible. Mostly deployed thermoset matrix systems which cannot be remolten
after curing process divine another aspect of maintenance issue. Recycling comes into
play here, too.
. standard aluminium solid rivets were replaced by Hi-Lokor Hi-Litefasteners for compos-
ite applications
3
: Introduction
. Thesis’ Structure
Composite materials have successfully arrived in aviation industry only at a rst glance.
In the next chapter, the state of the art in aircraft construction and composite technol-
ogy is presented nally outlining potentials and requirements via the introduction of
evaluation criteria. Chapter gives a detailed overview of fusion joining processes set-
ting spotlights on ultrasonic, resistance and induction welding. Subsequently, the eval-
uation of the latter three most promising process methods is carried out. Theoretical
background about heating models, mechanism and parameters is deduced in chapter 6
relevant for the experimental set-up and parametric study in the following two chapters.
The latter of both comprises a discussion of results. A transfer of these static results
into a continuous welding concept development is executed in chapter . Finally, the
conclusion provides a summary and outlook to next steps to be taken and leverage
points for future developments.
4
State of the Art, Potentials and Criteria
. Aircraft Structure
Main functions of the structure are to carry in-service loads, provide aerodynamic shape
and attachment for other systems as well as protect payload (PAX/ cargo) from envi-
ronment. Basically, the aircraft structure can be divided into primary and secondary
structure. Secondary structure represents all non-load bearing structures (e.g. brack-
ets, fairings, cowlings) rather for systems integration whereas the primar y structure can
be seen as skeleton of the aircraft which would endanger the aircraft upon failure. The
latter is a semi-monocoque structure, i.e. a combination of a load bearing (stressed)
thin sheet material skin with supporting stiffening (Figure .a). These are usually ()
stringers or longerons
, () circumferential frames, () skin and () clips (Figure .b).
(a) Semimonocoque Aircraft Structure
(FAA , -)
(b) Typical Aircraft Construction Joints
(Schulshenko , )
Figure . Primary Aircraft Structure
The use of materials has shifted in the last decades, as already indicated above.
Where over two-third of an aircraft was built out of aluminium and its alloys in the late
8s (Figure .a), the most recent aircraft developments consist predominantly of
composites and only of one-fth of aluminium and its alloys (Figure .b). A trend to
use more advanced lightweight metals and alloys like titanium, lithium and magnesium
is remarkable, too.
. “Longerons usually extend across several frame members and help the skin support pri-
mary bending loads. [...] These longitudinal members [Stringers] are typically more numerous
and lighter in weight than the longerons. (FAA , -)
2 State of the Art, Potentials and Criteria
68 %
 %
6 %
%
%
(a) A (8
*
)
 %
 %
 %
%
%
(b) A (
*
)
Aluminium Composites Titanium Steel Miscellaneous
Figure . Share of Structural Materials of Airbus Aircraft
(
*
Year of Maiden Flight; Data: Sturma 6)
. Composites Deployment
Composites are combinations of materials to combine favourable properties of at least
two different material types. In general, it can always be distinguished between ma-
trix and bres (Neitzel et al. , ). Fibres main task is to carry tension loads. With
the use of carbon, glass or aramid bres in aircraft manufacturing (Visakh and Lüftl
6, -), bres potentials are almost exhausted. Greater opportunity for improve-
ments provide matrix systems. Main tasks of the matrix is xation/positioning of -
bres in the compound, connection/transmission of forces among bres/laminate plies,
carrying transversal/shear loads, supporting bres under compression loads, acting as
crack stopper (ductile matrix systems) and protecting bres from environmental effects
(Schürmann , 8).
Despite some rare applications of other matrix types like metal
or ceramic
, “[p]olymer-
matrix brous composites, primarily because of their high specic strength and stiff-
ness, have found increasing application in aircraft structures. The critical need for sig-
nicant weight savings
, design exibility, and extended ight efciency for advanced
aircraft [...] has focused attention on composites [...]” (NRC 6, )
Structural observations in polymeric systems like order condition (amorphous or semi-
crystalline) or degree of cross-linking are decisive for mechanical properties and ther-
. MMC: Metal Matrix Composites
. CMC: Ceramic Matrix Composites
. However, MMCs with lightweight metals like magnesium, are reasonable in the aircraft
branch (Kaczmar et al. )
6
2 State of the Art, Potentials and Criteria
mo-mechanical behaviour of plastics. Based on that, it can be distinguished between
elastomers, thermosets (TS) and thermoplastics (TP).
What all have in common are acting binding forces namely primary and secondary
valence forces. Where in thermosets chemical covalent bonds (which are not weakened
until decomposition temperature) prevail, thermoplastic’s properties are dominated by
physical inter-molecular forces enabling a softening and melting behaviour hence higher
temperature dependency. However, natural limitation of usage temperature lies for all
plastics at about °C (Hopmann and Michaeli , ; Kaiser , 6-).
Figure . shows a schematic graph of the stiffness (Young’s Modulus) over tem-
perature for the three plastic classications. It stands to reason that TS and TP resin
systems are predominantly used due to their preferable thermo-mechanical behaviour
over a much wider range of temperature. Elastomers are rather characterised by their
high degree of elasticity leading to other elds of application, e.g. elastomer composites
for aircraft’s tire carcass (Mitchell and Landgraf 6).
It should be noted that due to partly crystalline structure, a second plateau can be
recognised between the glass transition point/area (T
g
) and the crystalline melting point
(T
SC
), which usually sets the service temperature range of semi-crystalline thermoplas-
tics. This difference is often emphasised by distinction of the terms melting for semi-crys-
talline and (gradual) softening for amorphous materials. For simplicity, both terms shall
equally describe the process of melting in the following.
80 40 0 40 80 120 160 200
10
0
10
1
10
2
10
3
10
4
T
g
T
g
T
sc
T
g
Temperature [°C]
Young’s Modulus [MPa]
Elastomer Amorph. TP
Semi-cryst. TP TS
Figure . Schematic Young’s Modulus over Temperatures for the Different Plastic Classica-
tions (Data: Domininghaus , 6; Bargel and Schulze , )
7
2 State of the Art, Potentials and Criteria
.. Thermosets (TS)
So far, thermosets are the most frequently used resin type in aviation composite appli-
cations. They show a better wetting of bres due to low viscosity caused by missing
cross-linking and low molecular weight before curing process starts. Curing process
can be interrupted actually slowed down by cooling, too, which is intensively used for
prepreg materials. Aforementioned, the net result of chemical curing is a clear solidi-
cation with increased strength, stiffness, creep resistance and glass transition/ melting
temperatures as well as good thermal and chemical resistance documented in much
long-term experience. Though, TS are rather brittle and due to their non-meltable na-
ture difcult to recycle (Strong 8, -6; Schürmann , 8). Eventually, melting
points can be found over the decomposition temperature (T
d
) hence the curing process
cannot be reversed any more. In addition, curing is characterised by intensive manual
preparation and long reaction times.
.. Thermoplatics (TP)
The increasing use of thermoplastic matrix systems is based on several advantages of
this resin type. Although Strong (8, 6) emphasises, the higher molecular weight
(higher viscosity) makes processing of composites more difcult, on the other hand,
it enhances most of physical and mechanical properties. Thermoplastic composites
(TPC) prot from the absence of complex chemical reactions, slow cure kinetics hence
shorter process times and no demanding cooled storage compared to TS. Among, TPCs
show enhanced toughness, innite shelf life, higher damage tolerance, fracture tough-
ness and impact resistance as well as good fatigue resistance. Non-ammability, better
environmental resistance against corrosion and solvents combined with very low level
of moisture uptake hence less degradation under hot/wet conditions seem to outweigh
the pros over the cons. Even of higher importance for following contemplations, TPCs
provide reprocessability and repairability contributing to a more cost-effective fabrica-
tion especially when regarding joining and recycling (Ahmed et al. 6; Yousefpour et
al. ; Costa et al. ). Nevertheless, disadvantages of high-performance thermo-
plastics can be named as requirements on high processing temperatures and pressure,
high raw-material costs and repair procedures are not well-engineered since thermo-
plastics have not found its way into aerospace applications broadly (Vodicka 6, ).
New developments have given rise to an innovative group of high-performance ther-
moplastics pushing the limits towards unprecedented melting points. In particular, the
8
2 State of the Art, Potentials and Criteria
group of semi-crystalline thermoplastics exhibit a unique character with better heat re-
sistance and higher strength properties of crystalline and more ductility of amorphous
regions (Menges et al. , ) which give them higher relevance. Especially the fam-
ily of polyetherketone (PEK
n
) have arisen particular interest of aircraft manufacturers
since they show superior proper ties concerning glass transition/melting temperatures
and strength/stiffness, receptively (Table ./Figure .).
Table . Mechanical Properties of Selected
Semi-Crystalline TP and TS
Semi-cr. TP Thermosets
PEEK PEK
UP VE EP
ρ g/cm
. . . . .
E GPa . .8 .
σ MPa   6 8 
ε % 6
T
g
*
°C -   
T
s
*
°C  6
HDT
*
°C   
8
*
T
g
: glass transition temperature, T
s
: solidus/melting
temperature, HDT: Heat Deection Test
Source: Schürmann , -;
Kaiser , 8, 8
0 1 2 3
100
200
300
400
PEKK
PEKEKK
PEK/PEEKK
PEEK
PEEEK
T
g
T
m
E/K
°C
Figure . Glass Transition/Melt-
ing Temperature of Dif-
ferent PEK
n
Derivatives
(Data: Domininghaus
, )
According to some industry experts, thermoplastic composites still have signicant
barriers to overcome before they are widely used in complex, contoured primary struc-
tures, particularly for aircraft produced in smaller volumes. These include cost, auto-
mated processing speed and quality, and lack of developed repair technologies. (Gar-
diner ) The price of PEEK is still an order of magnitude higher than average poly-
mer systems (Tooley , ). Jen et al. (8, ) report increasing applications
of APC- composites (CF/PEEK composites; USDoD , -) in high performance
aerospace structures due to superior mechanical properties. Such applications are the
undercarriage door of C- Hercules aircraft by Lockheed consisting of graphite/PEEK
thermoplastic composite, a composite ghter fuselage with a combination of various
thermoplastic prepreg materials including AS/PEEK (Vodicka 6, ) or the F-F land-
ing gear strut door and access panel, B- Bomber parts and Fokker- nose-wheel door
(Beland , 6-).
With the enhanced properties and thus opening of the market towards thermoplas-
tics, the composite world is changing drastically as many problems associated with the
9
2 State of the Art, Potentials and Criteria
classical thermoset matrix systems seem to be solvable now. The main drawbacks and
challenges are presented in the following.
. Main Drawbacks and Challenges
.. Fabrication
Composites are getting in a tight spot since material prices for aluminium and titanium
the most important lightweight metals in aircraft industry have dropped steeply in
the recent years (Tooley , ). Moreover, materials science develops continuously
new lightweight metals and alloys which provide high strength and low densities, too.
High standards in aerospace certication lead to even higher material costs by factor
two to three (Rao et al. ).
“Lockheed Aeronautical Systems Company have used thermoplastics in the manufac-
ture of an aircraft door structure [...] even though the raw material cost of the thermo-
plastic compared to aluminium was more than  times greater. Through automation
the cost of assembly was drastically reduced and the nal component was half the cost
of the aluminium equivalent. (Vodicka 6, )
Besides, fabrication processes of metal structures have become very efcient and
automatised in the last decades where manufacturing composites still requires a lot of
manual work and skilled labour.
For example, A side shells produced in the Augsburg plant are all made of prepreg/au-
toclave process. This production method promises the best and most accurate quality
of high performance composites paid with extensive expenses which can be identi-
ed as: (Irving and Soutis , )
energy demanding cooled storage of prepreg materials (limited lifetime)
need for high added-value processing materials (thrown away afterwards)
energy demanding curing in (argon/nitrogen) pressurised autoclave
costly non-destructive inspection (NDI) methods to guarantee high quality
labour-intensive process steps (stacking/lay-up/bagging/demoulding)
Shehab et al. (, ) provide an estimation of production time and manufacturing
cost for a single curved composite panel out of prepreg hand lay-up and autoclave cur-
ing process with ultrasonic NDT use case is reasonably comparable with an aircraft
shell. Just about two third of process time is directly related to the product (stacking,
:0
2 State of the Art, Potentials and Criteria
curing, nishing, testing), but almost  % are used for preparation or other process
steps. Regarding expenses, material costs still make up the largest part of overall man-
ufacturing costs, followed by labour costs with almost one fth due to many manual
steps and highly skilled workers. Automation and reduction of preparing steps would
lead to a better fabrication efciency.
In turn, Fokker (, ) lists advancements in thermoplastic fabrication as “more
freedom in dening form and in the number of production steps, [...] increase in speed
and simplicity”, as well as comparably easy realisation of integral structures instead of
one-shot necessity.
Integral structures are also the aim of
newest developments edging into the mar-
ket: additive layer manufacturing (ALM). Al-
though not ready for series production of
primary aircraft structures, this trend could
open a complete new way of engineering in
the near future as it offers an even higher
grade of tailoring and thus lightweight po-
tential in aerospace applications denitely
competing with composites. Analysing for
specic load paths, material is produced
only in position and direction needed us-
ing electron beam melting (EBM) or more
common selective laser melting (SLM) pro-
cesses, e.g. with titanium powder for high
strength properties (Figure .).
Figure . MG Bracket Rear (Löwer )
.. Joining
Classical metal construction in aviation branch is well known, huge experience and a
lot of data has been gained throughout the last  years. Manufacturing methods and
processes are rather easy as well as detectability and repairability during maintenance.
Based on that, aviation industry tried to implement composites very quickly by just
“replacing material” rather than developing a suitable manufacturing method. The con-
sequence is called “black metal design”. Carbon-bre reinforced plastics (CFRP) with
their characteristic black colour joined together with classical metal joining methods
like riveting or bolting.
::
2 State of the Art, Potentials and Criteria
Although the use of fasteners provides a robust process capable of high volume pro-
duction and joining of dissimilar materials combined with the ability to reopen and a
broad range of fastener types (Rotheiser , ), problems with mechanical fas-
tening are also well-known and reported repeatedly in literature (Ageorges et al. ;
Strong ; Schwartz ; Todd ; Qin et al. ):
Local stress concentrations at notches, drillings and cut-outs which cannot be
balanced due to lack of plasticity
Risk of delaminations originating during drilling holes
Different CTE
of fasteners and composite lead to thermal stresses
Enhanced corrosion and sealing issue, either by different galvanic potentials of
fastener and CFRP or by intruding uids (drainage areas/fuel) between fasteners
and composite
Destruction of physical and electrical continuity in brous composites
Expensive drilling tool demand and high labour skills required
Long process times for drilling operation ( holes per shell)
High number of fasteners required lead to additional weight
Regarding the latter, Vodicka reports a Lockheed investigation of an aircraft door
structure: “The aluminium door consisted of 6 parts and 6 fasteners while the ther-
moplastic equivalent used only  parts and  fasteners. [... T]he weight of the part
... was already 8 % lighter than the aluminium equivalent. (6, ) Integration and
assembly costs for an aircraft structure are estimated as - % of the nal aircraft
cost (Wedgewood and Hardy 6).
A more bre-fair approach and already widely spread is adhesive bonding. This me-
thod creates joints over large areas with uniform stress distribution, dissimilar mate-
rials, good fatigue behaviour whilst providing weight savings and incorporated sealing
(Rotheiser , -). Though, associated disadvantages are also renowned as in-
tensive surface preparation, difcult process control in industrial environment, long cur-
ing cycles, required skilled labour and no compatibility amongst some plastic materials
as well as with mass production benchmarks (Ahmed et al. 6; Schwartz ).
Their utilisation is based on the fact that these two methods are the only possible for
TS matrices. Moreover, fasteners offer a very reliable and predictable joining method
(Ageorges and Ye , 8). Therefore, main potential lies in the development of a -
. Coefcient of Thermal Expansion
:2
2 State of the Art, Potentials and Criteria
bre-fair, mass production compatible, reliable and predictable joining method desirably
without foreign objects.
.. Maintenance
Neglecting defects caused during manufacturing, in service life, an aircraft is exposed
to many different sources of damage. Therefore, certication authorities like EASA or
FAA introduced the condition of Airworthiness”
6
which is initially conferred. Continuing
airworthiness needs to be achieved and sustained through a systematic approach of
maintenance task and actions to minimise the risk of catastrophic failure caused by
manufacturing defects, corrosion, fatigue and accidental damage.
Not only, metals are more insensitive compared to composites and less prone to im-
pact damages due to elastic and plastic deformation. Moreover, composites are more
difcult to examine, detect and repair. Where scratches, buckling or folds in metallic
structure indicate clearly an area requiring repair (e.g. via doublers in unpressurised ar-
eas; Figure .6a), most defects in composites are hidden within the structure disguising
true scale of damage. Damage during service is most abundantly caused by low velocity
and rarely high velocity impacts. Matrix cracking and delamination invisible for naked
eye is then referred to as “barely visible impact damage” (BVID). There are several repair
approaches: ) patch repair applying a patch overlapping and bonding to the surface of
original laminate leading to a thicker structure and original strength. ) Taper sanded or
scarf repair are rather deployed to modern commercial aircraft. The detected damage
area is sanded circular (Figure .6b) in order to expose a section of each ply until the
damage is reached. By adding adhesive and overlapping plies (Figure .6c), thickness
is nearly remained, but a straighter and stronger load path is created. This technique
requires high skilled workers and due to its costlier method more time. Bolted repairs
in composites structures are rather counter-productive and thus seldom, but possible
and then quick, easy and heavier (R. A. Smith ; Zhang and Rong , -; FAA ,
-8-).
Thermoplastics and their ability to be remelted could ease maintenance tasks espe-
cially in terms of replacing damaged structure and reduce maintenance effort and time
drastically.
6. Air worthiness is the aircraft’s ability to y in safe conditions within allowable limits
:3
2 State of the Art, Potentials and Criteria
(a) Metallic Repair (b) Circular Sanding (c) Schematic Taper Repair
Figure .6 Aircraft Repair Methods (FAA , -, -6, -)
.. Recycling
Before major attention was drawn towards environment protection, composite waste
was disposed on landlls. The European Union established the “Directive on Landll of
Waste in order to reduce the amount of organic material landlled hence the prohibi-
tion to landll composites waste any more (EU ). Only one year later in , the
European Union introduced the “Directive on End-of Life Vehicle” (ELV) //EC re-
deeming the consequences of motor vehicles at the end of their lifetime causing eight
to nine tonnes waste per annum. The directive regulate by law a degree of recycling
of  % of initial vehicle weight from  on (EU ). The Process for Advanced
Management of End of Life of Aircraft” (PAMELA) project builds on the ELV directive for
the aviation sector. Although aircraft are made of materials that can be recycled or
reused in a number of ways, prior to PAMELA there were no standardised procedures.
PAMELA sought to ll this void, rstly by ensuring compliance with relevant waste reg-
ulation, and then, on a voluntary basis, by working towards achieving a target recycling
rate of 8 %, comparable to the EU End-of-Life Vehicles Directive (// EC), which
does not currently apply to aircraft. (EC ) As consequence, new joining technolo-
gies shall enable a later recycling and not introduce materials that cannot be recycled
any more. PAMELA reduced the non-recoverable material proportion out of a 6 tonnes
aircraft down to  %.
So far, with the use of predominantly (non-meltable) thermoset matrix systems, recy-
cling proved difcult. Mechanical processing (crushing) produces powdered or brous
recycling products still as mixtures of original materials. Fibre reclamation uses aggres-
sive thermal or chemical processes to separate bres from matrix and is due to their
thermal and chemical stability very suitable for carbon bres. Besides, thermal pro-
cesses are used in order to destruct the chemical cross-linking of thermosets. Most
widely spread is pyrolysis, a thermal decomposition of organic molecules in inert atmo-
:4
2 State of the Art, Potentials and Criteria
sphere. Also oxidation in uidised bed, i.e. combustion in an oxygen-rich environment
is eld-tested. Chemical approaches use reactive media like catalytic solutions, benzyl
alcohol or supercritical uids. Finally, combustion with energy and material utilisation
is still an even if undesired option, making use of the wastes caloric value when
burning (Pimenta and Pinho ; Pickering 6).
Thermoplastics could open a complete new eld for recycling. By simply reheating,
separation process can be eased drastically. Since no chemical reaction takes place,
the original state can be almost fully restored. “In this fashion, material usage efcien-
cies approaching  % can be reached. (Rotheiser , ) Either grinding/shredding
techniques for getting high-quality reinforcing material (press/injection moulding appli-
cations) or thermoforming processes for repurposing structures (Li and Englund ;
Schinner et al. 6).
:5
2 State of the Art, Potentials and Criteria
. Evaluation Criteria
Based on aforementioned drawbacks of state of the art technologies, evaluation criteria
shall be set in order to better exploit potentials of composites in terms of mechanical
performance, production efciency and bre-fairness design.
Furthermore, it is necessary to clarify the scope of relevance and priority of differ-
ent process related characteristics even before reections on available fusion bonding
technologies are made to set spotlights on important facets.
Besides fundamental evaluation criteria in industrial environment such as concerning
cycle time, costs, exibility or heat affected zone (HAZ), more attention shall be drawn
towards the driving criteria in this special case which are rated as primary evaluation
criteria (Figure .). These comprise in particular process capabilities and aircraft ap-
plicability with its respective derivatives and will be rated doubled in the following as-
sessment system contrary to secondary criteria. A short legitimation of investigated
criteria follows in the subsequent paragraphs.
.. Prerequisites
What should be pointed out is the fact that Chapter gives an overview of available
fusion bonding technologies in general. Not all of them do full given prerequisites.
Those will only round out the overview and present state of the art fusion technologies
while being excluded a priori from later decision process.
These prerequisites comprise bre-fairness (no bre interruption/distortion), longitu-
dinal areal lap joints (no butt joints), only use of a thermoplastic matrix system, main-
taining (aerodynamic) shapes and surfaces while access is only possible from one side
(adherend in tooling) and regards to large parts size.
.. Process Capability
Parameter. Seen as very crucial is the cycle time. It decisively inuences the productivity
hence guaranteeing a required production output (Sakamoto ). Exact quantication
has not been stated yet, so relative proportion will tip the balance.
Two others regarded are process pressure/forces and energy consumption. Both are
assigned to the sections Automation and Environmental Aspects, respectively.
:6
2 State of the Art, Potentials and Criteria
Automation. Since degree of automation in composite parts manufacturing of aircraft
industry is still relatively low compared to other branches namely automotive
, future
competitiveness in aircraft manufacturing will strongly depend upon the change from
manual to automated processes. Already in , US International Trade Commission
claimed that the “[...] higher level of production automation at Airbus contributes to
offsetting this [author’s note: although similar commonality in aircraft derivatives but
smaller production scales occurring] labor productivity disparity. (, -) Chursin
and Makarov (, -) assign a direct mathematical relation of automation to
competitiveness in their Quantitative Evaluation of the Firm Competitiveness.
Among the catchphrase Automation”, closed loop capability and robotic capability
are of special interest.
For an automated system in high quality production, a balanced and stable control cir-
cle is necessary. This should not only maintain the process but detect and correct pos-
sible disturbances. For this purpose, distinctive process parameters must be recorded
and processed. Also insensitivity to disturbances in general is important. There is a
cross-link to Reproducibility, too.
Robots due to their six axis construction are prone to high process forces, es-
pecially in the far distance eld. Reproducibility and repeatability suffer from harmful
effects caused by such high forces and moments. Furthermore, size and thus price is
dependent on weight and dimensions of the used end effector.
Process Chain Adoption. Process chain adoption represents the ability of a quick, sim-
ple and cheap implementation of new technologies in existing series production. Spe-
cial regards shall be made onto tooling, surface preparation and manufacturing schedule.
State of the art production consists of negative metal toolings for automated bre
placement (AFP) of UD plies (Figure .). Corrections or new acquisition of auxiliaries
are expensive and time-consuming hindering a desired quick implementation. Same
is valid for the manufacturing schedule. These procedures are highly developed and
fundamental changes can imply long periods until series production is back in desired
steady-state. Surface preparation partly picks up the automation target since most man-
ual steps shall be minimised as far as possible in order to improve productivity.
. BMW set new standards with an very high degree of automation in their i production re-
ducing cycle times by factor  (Schulze ). Amongst others, adhesive bonding technique
used in production is even fully automated (BMW Group ).
:7
2 State of the Art, Potentials and Criteria
Figure . Current Automated Fibre Placement Production Method (MTorres )
.. Aircraft Applicability
Aircraft manufacturing sets special standards due to presence of certication author-
ities. But also in terms of product size and tolerances, aviation industry is confronted
with other challenges than automotive or civil engineering.
Geometry. Geometry of parts exhibit probably the largest product dimensions
8
in engi-
neering and therefore transfer from laboratory scale to large scale joining and continuity
of joining process is very important. Since already Figure . implied, especially longitu-
dinal stiffener elements (longerons/stringers) are in charge of supporting the skin and
taking up primary bending loads over the whole length of the aircraft. A continuous joint
is therefore essential to enable an ideal force ow without too high local stress concen-
trations which would lead to additional local reinforcements hence more material and
weight always a crucial parameter in aviation. Only local joining methods, e.g. spot
welding, are eliminated self-evidently a priori by this prerequisite.
Lap joint design and accessibility is characterised by the scope of this work to form
longitudinal lap joints peculiarly of different section parts (upper/lower/side shells) to
assemble one barrel of aircraft section as well as longitudinal stringer joining over length
of a section. Airbus (6) promoted the so-called shells concept” (Figure .8a) with
associated advantage of weight savings due to solely lap joints needed, a lay-up and skin
thickness optimisation and individually tailored panels to local load cases. They made
another change in stringer design, too, according to new demands on joining technolo-
gies for CFRP structures. Since classical mechanical riveting was replaced by adhesive
bonding, conservatively shaped stringers (Figure .8b) were replaced by new compos-
ite omega shaped stringers (Figure .8c) which provide a greater surface important for
8. A- rear fuselage section (S6/8) side shell: length–  m, width– 6 m
:8
2 State of the Art, Potentials and Criteria
(a) Shells Concept
(Airbus 6)
(b) Conservative Stringer Shapes
(Niu and Niu , 8)
(c) A Composite Omega
Stringer (Luratec )
Figure .8 New Developments in A Structural Design
bonding lines and better inspectability. Both aforementioned changes are based on a
lap joint design or areal joining interface.
Performance. Peformance is generally characterised by static and dynamic consider-
ations. As static quantication, lap shear strength (LSS) if often referred to according
to test procedure ASTM D
. German and European test standards are provided by
DIN EN 6

, 

or 86

. Villegas et al. () regards fracture surface analysis
as supplementing measurement to LSS determination, since Guess and Allred ()
and Gleich et al. () showed both the dependency of joint geometry on shear and
peel stresses decisive for determined results.
Double cantilever beam (DCB) test is a standard test method to determine bond strengths.
Regarded is energy per unit plate width required to create a unit crack growth as a result
of peel forces perpendicular to crack plane. This values is referred to as G
Ic
. Testing
methods are described in ASTM D8-

or DIN EN 6

The simplicity of experi-
mental set-up, execution and evaluation with linear elastic fracture mechanics (Liu and
Gent , ) has it made used frequently in testing composite components.
Besides static considerations, especially in aviation structures, cyclic stresses are of
high importance. Therefore it is reasonable to execute a durability testing, too. Basis
for comparability gives an S-N curve or W diagram showing maximum stress am-
. ASTM D: “Standard Test Method for Apparent Shear Strength of Single-Lap-Joint Ad-
hesively Bonded Metal Specimens by Tension Loading (Metal-to-Metal)”
. DIN EN 6: “Determination of tensile lap-shear strength of bonded assemblies”
. DIN EN : “Non-metallic materials - Structural adhesives - Test method”
. DIN EN 86: “Determination of shear behaviour of structural bonds”
. ASTM D8-: “Standard Test Method for Mode I Interlaminar Fracture Toughness of
Unidirectional Fiber-Reinforced Polymer Matrix Composites”
. EN 6: “Carbon bre reinforced plastics Test method Determination of interlaminar
fracture toughness energy”
:9
2 State of the Art, Potentials and Criteria
plitude over achieved number of cycles until failure. Below a certain limit of applied
load called endurance limit no fatigue failure is expected. The procedure follows
ASTM D, too.
As aforementioned, fracture surface analysis gives additional information about fail-
ure modes and thus possible weak points in the joint. Detection and propagation of
failure types can be derived too with regard to certication and maintenance issues. Im-
portant criteria is the systematics or randomness of occurring failures in dependence
of the failure mode.
Certication. Certication authorities most noted representatives are in shape of
EASA (Europe), FAA (North America) and CASA (Australia) are responsible for con-
trol and execution of standards to guarantee aircraft’s ability to y in safe conditions
called airworthiness. By certifying the aircraft as well as its production, initial airworthi-
ness is conferred. Since these certication authorities have gained its experience and
ambition over decades of aviation and aviation accidents, certication framework has
become very strict and distinctive aiming at making ying as safe as possible. On the
other hand, in case of an accident, source of damage and failure should be identied in
any case to establish appropriate countermeasures.
For this purpose, full traceability of executed manufacturing steps and used material
batches must be enable. Process parameters, quality control and record of data go
hand in hand. Derivatives of this philosophy are requirements on reproducibility and
online inspection to maintain a degree of quality for initially checked production. If a
process exhibits high scatter, random failure occurrence or failure proneness just as
poor inspectability and less insight opportunities, certication is likely to be denied by
certication authorities.
Since used materials must be certied, too, in order to eliminate unknown materials
interactions, introduction of foreign materials is seen highly critical. Moreover, mate-
rial compounds can show effects like galvanic corrosion, poor adhesion or introduced
micro-cracks and notches causing stress concentrations which do most certainly in-
uence short-term and long-term static and dynamic material properties with unpre-
dictable consequences. Therefore, these cases must be avoided as possible.
20
2 State of the Art, Potentials and Criteria
.. Other Secondary Criteria
Investment. With every new technology and afliated acquisition costs, question of
investment and particularly of cost-benet ratio comes along. Cambridge Advanced
Learner’s Dictionary denes investment as “the act of putting money, effort, time, etc.
into something to make a prot or get an advantage, or the money, effort, time, etc.
used to do this. (CUP 8, 6)
It is difcult to quantify acquisition costs as they are depended on equipment, fea-
tures, purchase quantity, conditions and offers. An indication shall be given relatively by
equipment complexity and presence of recurring costs (RC).
Fibre-Fairness. One of the biggest disadvantages of conventional joining processes
like bolting or riveting is the absence of bre-fairness since load-carrying bres are cut
in order to drill rivet holes. Adhesive bonding follows the right approach which is picked
up by fusion bonding techniques in general. This criteria should only conrm this point
for regarded methods.
Heating Characteristics. Heating characteristics incorporate in this case basically two
points of consideration: rstly, the heat affected zone (HAZ) which usually refers to area
of heat induced. An important issue may be anisotropic behaviour of composites with
high thermal conductivities in and low thermal conductivity perpendicular bre direction.
Desired is a distinct heating area with a concentration of input energy in the respective
zone. Of equal importance is, secondly, the heating curve setting on the frame for maxi-
mum heating rates and thus cycle times. On the other hand, a too fast heating can either
cause material damage or overshooting temperatures above desired melting range. In
worst case, such an even short overheating can cause material decomposition and
therefore weakening of the whole component. A balance between heating speed and
safe temperature envelope must be found.
Maintenance. “Maintenance” is an important topic in aviation industry for two reasons:
rstly, for continuing airworthiness and thus permission to use an aircraft in commercial
service, a meticulous maintenance schedule is predened in order to detect, repair and
thus minimise the risk of catastrophic failure. Secondly, this goal turns out to be cru-
cial since the design service goal (DSG) desired life time an aircraft usually ranges
between  to  years. The importance of maintenance becomes obvious. In this con-
2:
2 State of the Art, Potentials and Criteria
text, question of detachability hence replaceability of damaged parts/structure plays a
major role as well as the por tability of maintenance equipment.
Environmental Aspects. In the backdrop of Clean Sky and positioning of this work in
the context of reduced CO
/gas emissions and noise levels, the ambition to follow this
mindset also during manufacturing is reasonable. The spotlight shall be turned to en-
ergy consumption of joining methods, responsible use of resources, presence of haz-
ardous materials as well as recycling during manufacturing and after reaching the end
of service life with a glance at legal framework presented in Section .. on Recycling.
22
Evaluation
Criteria
Primary
Aircraft
Applicability
Certication
Reprodu-
cibility
Online
Inspection
Foreign
Objects
Performance
Failure
Modes
Fatigue
DCB
LSS
Geometry
Tolerances
Lap Joint
Design
Large
Scale
Continuity
Process
Capability
Parameter
Cylce Time
Pressure
Energy
Cons.
Automation
Closed
Loop
Robot
Process
Chain
Adoption
Tooling
Surface
Schedule
Secondary
Investment
Complexity
RC Costs
Fibre-
Fairness
Heat
Affected
Zone
Maintenance
Environment
Energy
Cons.
Resources
Recycle-
ability
Figure . Primary and Secondary Evaluation Criteria
Overview on Fusion Joining Processes
In literature, many different classications and clusterings of joining technologies can
be found (Neitzel et al. , 6; Yousefpour et al. , ; Stokes 8, -;
Rudolf et al. , ). Commonly, they are divided into groups of mechanical fasten-
ing, adhesive and fusion bonding. The rst two can be assigned to conventional state
of the art technologies which shall be replaced.
Fusion bonding, also known as welding, describes according to DIN - a pro-
cess of joining two or more parts via creation of a material continuity under applica-
tion of heat and/or force with/without a ller (DIN - ). The advantage lies
in achieving bulk material strength in the welding joint and in its empirical values over
decades. Assembly can be achieved by heating thermoplastic material at a temperature
above melting point.
Generally, characteristics of heating classify fusion techniques. Detachability of the
created joint can play a role, too. In the considered case, both conventional types (me-
chanical fastening/adhesive bonding) produce non-detachable, whereas fusion bond-
ing provides also detachable joints. A modied overview and classication of Ageorges
et al. (, 8) can be found in Figure ., since he determined four classes of heat in-
troduction: bulk, two stage, frictional and electromagnetic heating. Within this clusters
several sometimes highly developed joining technologies are located.
Figure . Overview and Classication of Joining Techniques
Without prejudice, several studies state independently that the most promising ap-
proaches in fusion bonding technologies for composites are ultrasonic, resistance and
induction welding (Ageorges et al. ; Yousefpour et al. ; Ahmed et al. 6).
3 Overview on Fusion Joining Processes
In accordance to that, these three types will be elucidated more detailed than the brief
introductions of other techniques which shall only round out the listing. Final evaluation
shall be done regarding this three types only, too.
. Bulk Heating
Co-consolidation as a bulk heating typically resembles afliated bulk processes like
autoclaving, compression moulding or diaphragm forming. Basically, the entire part is
heated up to solidus temperature; thus, strength properties achieved usually equal par-
ent materials. Though, since pressure needs to be applied during the whole process
to prevent deconsolidation, complex tooling is necessary. Advantages are absence
of additional material hence no weight added and no surface preparation is needed
(Ageorges and Ye , ; Zhang and Rong , 6). Obviously, this process is rather
insufcient for large and complex parts (Davies et al. , ). Notwithstanding, it
was already used for McDonnel Douglas helicopter primary ight structure consisting
of CF-PEEK (Jouin et al. ), similar to the system envisaged in the regarded case.
For Hot-Melt Adhesives, a thermoplastic adhesive lm is inserted at the bondline in
molten state creating the bondline after solidication. This lm acts as shim between
two parts to be joined, too (Zhang and Rong , 6). Don et al. () showed the effect
of amorphous interface layers to lower scatter for strength properties. Both, Davies et
al. () and Fish et al. () investigated enhanced properties of APC- bondings
with PEEK hot melt adhesive lms.
Dual Resin Bonding, Amorphous Bonding, or Thermabond
®
processes consist of an
amorphous thermoplastic interlayer lm, co-moulded in a semi-crystalline thermoplas-
tic laminate prior to bonding. Compatibility at a molecular level of both, amorphous and
semi-crystalline matrix system, and thus good adhesion is essential for optimum prop-
erties. Already during laminating process, the interlayer lms are positioned at the later
areas to be bonded. Consolidation process creates a state where both materials are
present in molten state which is necessary to form the bond between both polymers.
After cooling, a thin layer of interlayer material remains on the surface. True joining is
eventually achieved by heating the interlayer over its glass transition temperature, but
below melting temperature of remaining composite structure (Smiley et al. , ;
Ageorges and Ye , ; Zhang and Rong , 6).
25
3 Overview on Fusion Joining Processes
For Thermabond
®
processes, a PEI
lm is combined with an APC- laminate.
This comes as ideal combination of mate-
rials since PEEK provides best mechanical
properties among thermoplastics and PEI
leaves a sufcient wide processing win-
dow preventing deterioration of structure
with a over °C lower melting point, dis-
played in Figure ..
Extensive studies were carried out on
dual resin joining. Davies et al. ()
found higher lap shear strengths for PEI
lms compared to PEI coated APC- lam-
inates.
0 100 200 300 400
0
1
2
3
4
Window
Process
PEI
PEEK
Temperature [°C]
Modulus [GPa]
Figure . Process Window of
Thermabond
®
Process (Data:
Smiley et al. , )
But Smiley et al. () point out the crucial correlation of preparation prior to co--
moulding and later achieved LSS. Very critical point for aircraft structures is the re-
ported reduction in strength for low temperatures of about –°C (Wu ). Besides,
amorphous interlayer showed degradation when exposed to solvents such as hydraulic
uid unlike hot-wet conditions (Davies et al. ; Smiley et al. ). Ageorges and
Ye (a) investigated resistance welding (Section ..) of thermoset-thermoplastic
compounds of epoxy composites with PEI lm.
. Two-stage Techniques
In two-stage procedures, heat is introduced by an external heat source which must be
removed before joining stage. Thus, size of joining parts is limited as a whole. Due to the
heating technique with highest temperatures below skin –, low thermal conductivity of
composites lead to long cycle times (up to  minutes for large parts). Moreover, high
pressures during heating and particularly joining ensure good consolidation, yet, can
cause warpage/ow in hotter inner regions of the laminate. Another main disadvantage
is the possibility of contaminations and inclusions (Ageorges et al. ).
One of the most popular approaches within this group is Hot Plate Welding due to
its simple, robust and efcient process with strong welds. A crucial process parameter
is the pressure parts to be joined are pressed against the hot tool since, if too little, no
good heat conduction is achieved and if too high, melt may be pushed out of the bondline
(precaution taken with mechanical stops). To prevent molten plastic from sticking to
the tool, it is usually coated with PTFE. A classical hot plate welding process consists
26
3 Overview on Fusion Joining Processes
of four phases: heating, tool removal, joining and consolidation (Figure .a) Typical
applications are in automotive and infrastructure sector, e.g. joining of large-diameter
plastic pipes (PDL 8, -; Stokes 8).
Hot Gas Welding and Extrusion Welding uses hot gas (instead open ame as for
metals) for heating bond surfaces forming a groove for pushing in a ller rod, respec-
tively for extrusion welding molten ller material is extruded into the joint. This pro-
cess shows good automation possibility albeit mostly applied manually. Unless high
exibility in part geometry and complexity just as portability of equipment, it is a slow
and difcult to control process hence not suitable for high-volume production (Yousef-
pour et al. ; Stokes 8).
Infrared Welding (IRW ) is rather new and a non-contact joining method within this
subgroup using high-intensity quartz lamps emitting intense IR radiation to heat the
bonding surfaces (Figure .b). Advantages are high heating rates, reduced contam-
ination risk hence higher reproducibility with low scatter strength properties (good for
certication issues) as well as exibility to join large at or curved parts and good au-
tomation potential with online inspection. Disadvantages can be noted as strong depen-
dence of the heating process on the colour of parts inuencing absorption characteris-
tics. This irregularities can lead to surface overheating or deconsolidation and warpage
for deep heat penetration (Yousefpour et al. ; Costa et al. ).
(a) Schematic Phases of Hot-Tool Welding (b) Schematic Infrared Heating
Figure . Two-Stage Techniques (Costa et al. , 6-6)
Laser Beam Welding (LBW) or Laser Transmission Welding (LTW) in context of ther-
moplastic composites is also a rather new and not deeply investigated non-contact tech-
nology but promising in terms of suitability especially for thin to medium thick parts in
aerospace applications. “Moreover, size, geometrical requirements, and specications
of aeronautical parts can be more easily fullled by applying LTW technique rather than
alternative technologies. (Labeas et al. , ) Notwithstanding, one of the joining
parts (top materials) needs to be transparent in order to guarantee absorption of laser
27
3 Overview on Fusion Joining Processes
energy by the bottom material and simultaneous heating of both parts via conduction.
Fibre reinforced composites are therefore eliminated for this technique. Unless “trans-
parent” set-up, only butt joints are within possible applications of LBW. For this process,
the laser beam decomposes polymeric material while a thin molten layer remains which
is used to form the bondline after solidication under pressure. Either way, it exhibits low
cycle times, clean process leading to weight and cost reduction while achieving accept-
able strength. Drawbacks are transparency prerequisites for lap joints and laser inten-
sity as crucial process parameter which is decisive for a successful heating rather than
polymer decomposition. Most abundantly utilised in polymer welding, a laser-assisted
thermoplastic composite tape/tow winding process has been developed in composites
eld (Yousefpour et al. ; Costa et al. ; Labeas et al. ).
. Frictional Heating
According to the Handbook of Plastics Joining, frictional welding is “a welding method
for thermoplastics in which friction provides the heat necessary to melt the parts a
the joint interface. (PDL 8, ) Therefore, frictional movement classies different
heating principles.
.. Various Processes
Spin Welding was already introduced in the s and is still one of the most common
friction welding techniques. Circular shaped parts whereof one is xed and one is
rotating about its axis with a dened angular velocity are pushed together under a
specic axial pressure. At the interface, rubbing due to relative movement induces fric-
tional heat that eventually melts the polymer. After retarding, solidication forms upon
cooling the bondline. The whole process can be described in four phases: ) initial heat-
ing, ) un-steady melting and ow, ) steady state ow and ) solidication. Main pro-
cesses parameters are, partly aforementioned, angular velocity, welding pressure, forg-
ing pressure and welding time. For an optimum welding pressure and angular velocity,
an optimum can be recognised, whereas for longer welding times a better strength can
be achieved. Advantageous is the simplicity, little to no surface preparation, high weld
quality with good reproducibility and a good automation of drill presses or lathes. Main
disadvantages are restrictions to circular shaped parts; otherwise a angular alignment
must be applied or orbital welding is getting introduced. Thereby an oscillatory motion
forces each point on a small-radius circular curve. Best applicable parts are thermo-
28
3 Overview on Fusion Joining Processes
plastic composite tubes welded together or to at panels (Ageorges and Ye , ;
Yousefpour et al. ; Stokes 8).
Unlike the latter method, Linear Vibration Welding creates heat via an oscillatory
movement on a linear path, but the principle of spin welding remains the same. Two
parts pressed together and with a relative motion to each other in an appropriate fre-
quency induce heat by Coulomb friction and shear stresses. Due to higher rigidity of bre
reinforced composites, higher frictional energy leads to shorter heating times. Further
inuence on strength has the bre orientation, welding pressure, weld type and most im-
portant penetration depth. Moreover, this process is applicable to various thermoplas-
tic types of small-to-medium size (at-seamed) without intensive surface preparation.
Drawbacks are inability to weld non-at-seamed parts, low modulus thermoplastics and
particularly for composites joining, bre distortion and displacement is very likely (PDL
8, -; Stokes 8).
Friction Stir Welding (FSW) is rather for particle-lled or short-bre reinforced plas-
tic since a metallic head-pin intrudes with a rotating motion in the butt weld between two
closely positioned parts. The shoulder of the tool eventually sets on the surface of both
parts and the heat induced softens the thermoplastic. While the tool moves along the
bondline, both parts are welded together. Notwithstanding, greatest disadvantages are
bre breakage due to the penetrating tool and the restriction to butt welds (Yousefpour
et al. ).
.. Ultrasonic Welding
Application of ultrasonic vibrations in aviation manufacturing goes far back in history:
since over a hundred years, ultrasonic waves have been used as non-destructive defect
detection method and are still state of the art to inspect composite aircraft parts these
days (Rose et al. ).
The principle of Ultrasonic Welding (US) lies in oscillatory vibrations at high frequen-
cies about  to  kHz (Costa et al. ) and low amplitudes of  to 6 µm (Stokes
8) which act perpendicular to the faces to be joined (main process). Surface and
intermolecular friction induces heat utilised to soften the thermoplastic system. Volkov
et al. (a; b) showed the inuence of surface micro irregularities on later weld-
ing quality. This effect can be used when man-made asperities are introduced called
energy directors (ED) or susceptors (Ageorges et al. ). Once these susceptors are
molten, they soften the interface with a subsequent diffusion/entanglement of polymer
chains forming the bondline. However, a second process produces shear joints made
29
3 Overview on Fusion Joining Processes
without any articial susceptors since a major por tion of vibration energy is converted
directly to frictional (shear forces) (Figure .a; Yousefpour et al. ).
Equipment essentially consists always of seven components: stand, generator/power
supply, converter, booster, horn/sonotrode, xture and controls (Rotheiser , ).
A schematic set-up is shown in Figure .b. The power supply provides high frequency
electrical energy exciting a piezoelectric or magneto-restrictive material as part of the
actuator. This hosts the converter (converting electrical energy in vibrations), booster
and horn whereof the latter two increase amplitude of originally very small converter
oscillations nally to  6 µm for amorphous and up to  µm for semi-crystalline
materials and connect actuator to the top surface of the substrate.
“The objective is to focus the heat at the bondline and keep the remainder of the part
interior and the heat-affected zone, or HAZ (the area next to the bond) from deconsoli-
dating. Since this method only heats the parts in the area being bonded, pressure only
needs to be maintained directly under the sonotrode and any associated HAZs. Pres-
sure between  and  kPa ( and  psi) must be maintained until the part has
cooled to below the T
g
of the matrix resin. (McCarville and Schaefer , )
(a) Joint Variants (b) Schematic Welding Devices
Figure . Ultrasonic Welding Fundamentals (Yousefpour et al. , 6)
Benatar and Gutowski (8) cluster ultrasonic welding into ve steps: ) mechanics
and vibrations of the parts, ) viscoelastic heating of the thermoplastic, ) heat transfer,
) ow and wetting and ) intermolecular diffusion.
Benatar and Gutowski (8) further pointed out the importance of detecting dynamic
mechanical impedance as relation to molten polymeric material ow for online monitor-
ing of weld quality. For ow improvement and gap lling, typically, a thin layer of neat
resin lm is inserted (Campbell , ). Villegas () introduced in-situ monitoring
of ultrasonic welding through power-displacement curves recorded by microprocessor--
controlled machines itself.
30
3 Overview on Fusion Joining Processes
Process parameter are reported as welding pressure, welding amplitude, welding time
and welding frequency. Benatar and Cheng (8) and Strong et al. () investigated
them with concluding weld strength (quality) improves with higher pressure (better en-
ergy transfer), longer welding time and increased amplitudes (energy dissipation in-
crease), whereby an optimum can be observed over that a degradation of property val-
ues sets in again. Suresh et al. () report a similar behaviour for inter face temper-
ature and ultimate tensile strength of ABS specimen showing an optimum at a certain
temperature, too, for both, amorphous and semi-crystalline. In addition, they regard vis-
coelastic heating as most crucial depending on applied frequency, square of the ampli-
tude and loss modulus of material . The heat generation concentrates around asperities
in the surface (Wolcott 8).
Volkov and Kholopov (8a; 8b) found induced misorientation and distrotion of
bres due to applied welding pressure and proposed a reduced (not removed to main-
tain physical/acoustical contact) pressure at thermoplastic’s melting point and a re-in-
crease after vibration phase to guarantee a good consolidation. Fischer et al. ()
investigated detrimental overheating and penetration of the horn into the laminate as
well as disruption of bres depending on process parameter .
Suresh et al. () pointed also out
the good welding quality of amorphous as
well as semi-crystalline thermoplastics in
the near eld, but rather poor quality for
semi-crystalline thermoplastics in the far
eld. This phenomenon does not occur
for composites due to reinforced stiffness
(Grimm ).
Moreover, semi-crystalline thermoplas-
tics need more energy since they absorb
more energy and additional energy is re-
quired to break crystalline structured ar-
eas and thus, they are more difcult to
control as the process window for tem-
peratures is narrower as depicted in Fig-
ure . (Rotheiser , 8-8).
T
g
T
m
amorphous
semi-cryst.
Temp.
Specic Heat
Figure . Specic Heat over Temperature
for Thermoplastic Types (Data:
Rotheiser , 8)
The inuence of energy directors was and still is subject of intensive studies since
they are source of heat and therefore crucial for a successful process. Where ED mod-
elling is achieved quite easily in plastics manufacturing, composites face serious prob-
3:
3 Overview on Fusion Joining Processes
lems since bres must not be distorted. Furthermore, an additional process step would
be necessary. Triangular protrusions are most commonly used although other shapes
were investigated (Benatar and Gutowski 8) and tie layers with modied and prefer-
ential melting properties were proposed by Tateishi et al. (8) and Zach et al. (8)
instead of any energy susceptors. Most recently, investigations go in the direction of at
energy directors in form of lms (Senders 6; Villegas and Palardy ; Palardy and
Villegas ; Villegas et al. ). These at EDs turn out as competitive alternatives to
standard triangular shapes. Although the latter possess superior properties compared
to semi-circular or rectangular shapes (Suresh et al. ), differences in weld strength,
dissipated power or heating time compared to at EDs are rather marginal (Villegas et
al. ).
Stokes (8) introduces the issue of joint positioning and design in ultrasonic weld-
ing since parts geometry is decisive for vibration behaviour and thus transmission of
energy to the joint inuencing heating and melting.
Strong et al. () determined four failure modes for ultrasonic welded lap shear
joints (in order of increasing strength): ) weld interfacial failure in resin-rich areas,
) combined interlaminar and interfacial failure, ) interlaminar failure above and be-
low energy susceptor layer and ) coupon failure due to bre damage (no weld failure).
Rotheiser (, 8-8) lists advantages amongst others:
No additional materials hence inherently lower in cost and less expensive to dis-
assemble for recycling
Ease of assembly– requires only alignment of two parts in xture thus it is well
suited to automated assembly
Permanence– permanent joint without creep, cold ow or stress relaxation ef-
fects, or other environmental limitations
Contour freedom– no limitations in contour geometry like f.i. spin welding
Hermetic seals possible
Energy efciency– highly energy efcient process including no excess heat which
must be removed from workplace
Clean Atmosphere– Unlike adhesive and solvent joining systems no environmen-
tal requirements
Immediate handling– no curing times delaying process chain hence good for au-
tomated assembly line applications
High production rates– due to very short cycle times  to 6 parts per minute
are possible
32
3 Overview on Fusion Joining Processes
Very short cycle times hence very small heat-affected zones (Todd ). Krüger
et al. () showed also the opportunity of joining combined composite-metal com-
pounds (AlMg
) with ultrasonic welding process achieving satisfactory tensile/shear
properties. Nonetheless, HAZ can be slightly increased with a greater sonotrode diame-
ter to lower applied welding pressure avoiding disruption of bres (Volkov and Kholopov
8a, 8b).
In maintenance and repair, ultrasonic processes are well-known. Damaged areas are
drilled out and lled with a plug of thermoplastic resin. Only for small areas, reduc-
tion in strength due to cutting of load carrying bres is acceptable (Vodicka 6, ).
Concerning possible maintenance applications, ultrasonic welding equipment has been
characterised as “[...] too heavy for practical in-eld work. (Lewis )
Schwartz () sees the main barrier for ultrasonic welding applications to continu-
ous-bre reinforced materials at the insertion of energy susceptor on sheet components
with associated possibility of bre disruption under high oscillation motion. Another is-
sue is heat conduction by carbon bres away from the bonding interface consequently
increasing cycle time (C. Eveno and Gillespie 8). Size and power of welding machines
limits the size of a one-shot bonding area (Benatar and Gutowski 86). Largest ultra-
sonic welders cover areas of about . m × . m (Rotheiser , 8). Otherwise, a
multiple horn application must be deployed (Taylor and Jones ).
Frantz () emphasised the limitation of ultrasonic welding to large and complex
shaped parts whereas McCarville and Schaefer (, ) rate scaling of an ultra-
sonic welding process to large and/or complex shaped parts with reference to Davies
et al. () at least as critical. Heimerdinger () reported unsuitability of ultrasonic
welding for repairing large parts as bonds produced showed only poor strength, only
partial bonding or burnt material.“The ultrasonic energy is highly directional and multi-
ple passes of the beam are required to cover a large area. (Vodicka 6, )
Generally, ultrasonic welding is only applicable to (nearly) at joining surfaces large
enough to move over with the sonotrode. Further disadvantages are tight surface toler-
ances and prerequisites on prior manufacturing processes which can achieve them or
even enable insertion of energy directors. In addition, electrical components near to a
part under repair, e.g. during maintenance, can be damaged due to vibrations. Workers
must be protected from the shrill, high-pitched whistlelike sound created by the horn.
Basically, the simplest equipment versions are not very expensive, but with additional
accessory to improve weld quality and ensure stable process, price can increase rapidly
33
3 Overview on Fusion Joining Processes
(Rotheiser , 8-8). Moreover, with such a method, only non-detachable joints can
be manufactured akin to its type of metal joining.
“This process is somewhat similar to spot welding[...]” (Campbell , ), for
which Yousefpour et al. () sees great potential especially in the aerospace branch.
Nevertheless, in context of this work, a continuous joining method is sought for. Inves-
tigations in direction of continuous ultrasonic composites welding can be found inten-
sively upcoming in the last decades (Villegas and Bersee ). Joining of woven and
non-woven as well as large parts, sheets and plates have been reported with continu-
ous, scan or sequential ultrasonic welding processes, respectively (Grewell et al. ;
A. Benatar et al. ; Benatar and Gutowski 86; Lu et al. ).
As depicted in Figure .6, either the adherends are positioned on a rotary drum or
the horn moves over stationary clamped parts; alternatively, a moveable table positions
parts under the xed horn. The former is normally featured with a constant feed, gap and
a round edged sonotrode preventing snagging and better pressure distribution. During
scanning ultrasonic welding, sonotrode traverses or scans the pars at a lower veloc-
ity hence longer ultrasonic vibration times due to thicker parts joined with this variant
(Gallego-Juárez and Graff , -).
(a) continous
(b) scanning
(c) sequential
Figure .6 Variants of Ultrasonic Welding (Gallego-Juárez and Graff , -)
Most of these methods are used to join exible thermoplastic lms, woven and non--
woven fabrics and coated materials (BRANSON ). Particularly for neat thermoplas-
tics, it is the most promising method achieving high quality and reliable weld seams
(Khmelev et al. ).
34
3 Overview on Fusion Joining Processes
Those principles can be modied to ul-
trasonic seam (roll) welding (Figure .)
where circular disk sonotrodes on trans-
ducers are moved translationally for gen-
erating vibrations suitable for long length
sheet joining particularly of dissimilar ma-
terials. An alike process is ultrasonic
torsion (ring) welding which is however
rather for spot welding of circular joint
shapes (Ahmed , -).
Figure . Ultrasonic Seam Welder (Ahmed
, )
Typical applications are within various branches, e.g. electrical, computer, automo-
tive, energy, medical and packaging and aerospace engineering is keen on exploiting
this technology for lightweight applications (Yousefpour et al. ; Costa et al. ).
Particularly optimum process parameters and control in a stable manner is the goal of
several studies (Siddiq and Ghassemieh 8; Krüger et al. ).
As already implied before, promising studies have been carried out to expedite the
continuous approach of ultrasonic welding for composites, too. Already in , EADS
conducted experiments nally leading to a patent in a “Ultrasonic Assembly Method”
(Soccard ) providing a continuous process. Levy et al. report that “[u]ltrasonic con-
tinuous welding of thermoplastic composite plates is a very promising process of partic-
ular interest for the assembly of aeronautics large parts. [...] First results reveal a good
mechanical quality of the welding. In particular the advance of the sonotrode enable air
removal along the director and avoids the trapping of bubbles. [...] This opens possibil-
ities for this process to be used at an industrial level to assemble large parts keeping
an excellent weld quality. () Fokker Aerostructures realised the A8 xed leading
edge with ultrasonic spot welding and Boikon expanded this approach to a ultrasonic
UD tape laying head establishing a new xed wing leading edge concept (Figure .8).
“With ultrasonic bre placement, the challenge was to succeed in tacking a tape onto
an underlying thick stack of plies at high speed, without slowing down the process and
thus making it ineffective. (Offringa )
Back from manufacturing to joining, the Faculty of Aerospace Engineering of Delft Uni-
versity of Technology developed under the F Eco-Design programme spot and sequen-
tially ultrasonically welded aircraft primary structure parts consisting of CF/PEEK and
thus proved the feasibility of this joining method for large-scale applications (Palardy et
al. ). As one outcome, Senders (6) developed and proved in his master thesis
the feasibility of such a technology. In an annexed article, Senders et al. introduced a
35
3 Overview on Fusion Joining Processes
(a) Schematic (b) Development Robotic Cell
Figure .8 Ultrasonic Fibre Placement Head and its Application (Offringa )
“Zeroow” approach inheriting “[..] a welding procedure in which no squeeze ow of the
energy director is required to achieve sufcient weld strength, [and which] is enabled
by the use of very thin (.8 mm) at energy directors. The zero-ow approach allows
continuous welding of stiff thermoplastic composite plates since it does not require
local deformation of the adherends as the sonotrode moves along the weld line. The
results presented in this paper for a basic zero-ow continuous ultrasonic welding pro-
cess prove its feasibility and indicate its potential to deliver high-strength welded seams
at very high speed. (6)
The ultrasonic welding methods was favoured for joining the - Westland heli-
copter tailplane (horizontal stabiliser and n) unlike resistance and induction technolo-
gies (Cole ).
. Electro-Magnetic Heating
As the name indicates, heating is induced via electro-magnetic elds, waves or effects.
The appearance of heat further depends on the kind of physical principle used out of
the huge eld of electro-magnetic effects. For industrial process, four methods are most
common: microwave and dielectric heating as well as induction and resistance welding.
.. Various Processes
Both, Microwave and Dielectric Heating are underprivileged a priori when it comes
to “... multi-layer composites, which are excellent shield in the microwave range, mi-
crowave welding is poorly suitable especially when composites are reinforced by car-
bon bers.”(Boyard 6, 6) As a consequence, bulk heating sets in within the top
36
3 Overview on Fusion Joining Processes
layers (Volpe 8); same holds true for dielectric heating (Benatar and Gutowski 88).
Nonetheless, by adding suitable absorbent materials in the interface, Varadan and Varadan
() achieved good welding qualities and applications for glass or aramid reinforced
composites are possible anyway (Vodicka 6), but out of scope in this work.
Microwave welding came up with development of magnetron in the s (Costa et
al. ). Generally, a range of x 
8
to x 

Hz is referred to as microwave radi-
ation. Within this wavelength range, most of thermoplastic systems do not show any
signicant heating when exposed to radiation. In turn, microwave energy is absorbed
by microwave sensitive implants causing heating which leads through conduction to
a melting of adjacent polymeric material (Wise ). The physical principle of mi-
crowave heating is a combination of several loss mechanisms, amongst others dipolar,
Maxwell-Wagner and Ohmic loss effects which are often summarised in an effective
loss factor (Metaxas and Meredith 8). The four process steps are ) heat genera-
tion, ) heat conduction and melting, ) ow and diffusion and ) cooling. For this, main
process parameters inuencing the amount of heat generated can be characterised as
heating time, power level, welding pressure and percentage of conductive polymer (PDL
8, 8). Advantages are freedom for designers since parts are usually not excited by
radiation, possibility for complex three-dimensional joints in very short welding times
of less than a minute, fast and clean behaviour, automation in a continuous process,
especially for butt welds (Wise ; Yousefpour et al. ) and consequently high
processing speed and energy efciency (Ku ). A wide range of implant materials
is available, ranging from metals, carbon to polymers all are consumable (Costa et al.
). Although these implants act as very local heat sources, quite a uniform heating
is achieved; high energy coupling efciency causes volumetrically heating (Yarlagadda
and Chai 8; Ku ).
However, the size of joints is limited
with proportion of absorbent material.
Figure . shows a schematic continu-
ous butt welding process with microwave
heating. A disassembly can be consid-
ered with a similar process of reheating
but inverse “pressure to pull the parts
apart (Yousefpour et al. ).
Figure . Schematic Continuous Mi-
crowave Joining Set-Up (Siores
and Rego )
37
3 Overview on Fusion Joining Processes
Where materials with low or medium dielectric loss factor require no electromagnetic
absorbent material for microwave joining, joint elements with high dielectric loss factors
are rather predestined for dielectric heating. The principle is the same as for microwave
radiation, albeit the polymer is directly heated at much lower frequencies.
Dielectric or Radio Frequency Welding (RF) uses  to  MHz radiation (PDL
8, ). Key process parameter for good weld qualities is undoubtedly the dielectric
loss factor which is decisive for heat amount induced during irradiation. Thus, PVC, PU
or PA are favourable joining materials for this process. Further inuences have the di-
electric power source, thickness, area and properties of welded parts as well as welding
time and pressure. Just as for microwave heating, the dielectric effect is enhanced for
reduced part or increased weldline thickness (Yousefpour et al. ). Process advan-
tages are simple and compact set-up which does not introduce other bonding materials
minimising risk of inclusions (PDL 8, 8). However, for composites, main drawback
is the likelihood of bulk heating hence de-consolidation which requires counter-mea-
surements like expensive tooling. Corrective can be achieved by adding thin layers of
higher dielectric loss material than the surrounding composite. As result, the thin layer
melts before deconsolidation of the entire composite part.
As depicted in Figure ., contacting electrodes directly transmit the alternating eld
onto the joining elements, which are usually automated bonding lm and thin sheets
bonding processes (Stokes 8). Another application is thermoset or adhesive curing,
too (Yousefpour et al. ).
Figure . Schematic Dielectric Joining Set-Up (Yousefpour et al. )
38
3 Overview on Fusion Joining Processes
.. Resistance Welding
Resistance Welding (RW) often
also referred to as Resistive Implant
Welding or Electrical-Resistance Fusion
(Ageorges et al. ) uses the principle
of a current-carrying conductor (heating
element) inserted in a sandwich congu-
ration producing heat directly at the bond-
ing interface due to resistive effects (Fig-
ure .).
Figure . Working Principle of Resistance
Welding (Hou et al. b)
Once the current is applied and heat losses from the joint are exceeded, supplied
heat energy conducts to adjacent thermoplastic structure which eventually reaches its
melting point. The longer the energy supply persists, the deeper is the heat penetra-
tion. In order to avoid unnecessary distortion of bres, heat-affected zone is tried to be
minimised to the bonding layer. After satisfying heating, the current is took away and
cooling sets in under application of pressure to guarantee a good consolidation with
sufcient intimate contact and molecular diffusion. The heating element remains in the
bondline and can be used for reheating at a later point in time, e.g. for replacements
during maintenance. Instead of added implant material, electrical conductive carbon
bres open the possibility to be used as “internal” heating element (Stavrov and Bersee
; Hou et al. b; Yousefpour et al. ).
This process features a rather sim-
ple set-up with non-expensive and easily
portable equipment (Yousefpour 6).
Most of the parts are standard parts like
implant mesh plies, pressure application
tools, electrical power supply, clamping
and measurement devices (Figure .).
Welding pressure should have a uniform
manner and can be applied diversely,
e.g by pneumatic cylinders or calibrated
spring clamps. Electrical power supply
can either provide direct or alternating cur-
rent (Stavrov and Bersee ).
Figure . Resistance Welding Set-up and
Process Steps (PDL 8, 8)
RW can be divided into three simple process steps: ) assembly of the stack-up,
) welding (heating) and ) cooling. According to that, three stages can be observed:
39
3 Overview on Fusion Joining Processes
) ramp up, ) peak current and ) ramp down stage with an overall process time of
typically one minute (PDL 8, 8-8).
Various process parameters can be identied, starting with the energy (current) ap-
plied. According to Ohms and Joules Law, energy dissipated can be derived as in (Eq. .)
from the material dependent resistivity used for the conductor.
E = U ·I ·t = R ·I
2
·t with R = ρ ·
L
A
or R = γ ·
L
w
(Eq. .)
with electrical current I, time-dependent resistance R, (material) specic resistance ρ
anc conductor’s length L and cross-sectional area A. M. Hou and Friedrich () in-
troduced the specic resistance γ as proportionality factor of length L and constantly
assumed width w of the conductor. Since a huge temperature range is expected, the
change of electrical resistance over temperature must not be neglected. The correction
follows the linearised Taylor series
R(T ) = R
0
·[1 + α
T
0
·(T T
0
)] (Eq. .)
with the resistance-temperature coefcient α
T
0
valid for a starting temperature T
0
. For
typical conductor materials (e.g. copper, aluminium), α
T
0
> 0 holds true hence an in-
crease of resistance and heat energy for higher temperatures. Investigations on this
phenomenon were carried out showing an increase of stainless steel resistance of about
 % and a decrease of carbon bre heating element resistance (with rather insulating
behaviour) between 6. and 6 %, both at °C (Stavrov et al. ; Ageorges et al.
a).
Proleptically, material with its resistivity and temperature behaviour in combination
with the current applied forms the rst process parameter.
Furthermore, there are several possible congurations for welding execution: ) con-
stant input voltage during entire welding (but variable power level due to changing elec-
trical resistance over temperature), ) constant power during entire welding process
hence temperature peaks which a difcult to predict, ) constant temperature which is
maintained via thermal sensors to monitor, control and adjust power, ) ramped voltage
with a gradually increase, constant heating rate for a more uniform temperature distri-
bution and less heat losses in surrounding medium hence lower input energy, or ) im-
pulsive resistance welding with energy supplied in form of intense pulses (followed by a
to seconds interruption) leading to less heat losses hence less energy consumption.
Although others consistently show more evenly temperature distributions, the constant
40
3 Overview on Fusion Joining Processes
power approach is more practicable and thus more abundantly used (Yousefpour et al.
; Arias and Ziegmann 6; Ageorges and Ye b; Stavrov and Bersee ).
Monitoring and controlling of electrical resistance/input power as well as associated
heating and cooling rates are decisive for later weld quality and thus mechanical prop-
erties of the joint (Yousefpour et al. ).
As addition to (Eq. .), the welding power/energy per unit area [
W
m
2
,
kJ
m
2
] can be com-
puted according to
P =
R ·I
2
L ·w
and E = P ·t =
R ·I
2
L ·w
·t (Eq. .)
Interestingly, Hou et al. () proved a difference in processes with high power levels
and short welding time to a process vice versa. Tests revealed for increasing power
levels an decrease in lap shear strength as well as a dramatic reduction of LSS for lower
welding energies.
On the one hand, a good thermal insulation and carefully dimensioned input energy
(Costa et al. ) can increase weld quality, but on the other hand, the power supply can
also be a limiting factor for power level hence weld time and size (Stavrov and Bersee
). Thus, the insulation issue is closely intertwined and turned out to be crucial
since satisfying weld results are not achievable without it (Xiao et al. ; Jakobsen
et al. 8; Stavrov and Bersee ). Either tooling is heated in a way the bondline
does not reach melting temperature or the other way around, bulk heating sets in with a
consequent deconsolidation. Various different insulation material like ceramics, wood,
silicon rubber or brous matter were investigated for this reason also with additional
coating foils (e.g. PTFE) to prevent sticking and improve surface quality. The selection
of material must be carefully done since the implant remains in the bond and compati-
bility with joint materials is required for long durability. In literature, only stainless steel
and carbon bres (CF) co-moulded in several resin layers are reported as heating ele-
ment materials whereof carbon bre show very good compatibility with the prevailing
structure as it is produced from the same source. Metal mesh are advantageous in
terms of better performance, consistency and strength of formed joint as well as less
sensitivity to process disturbances. However, issues like added weight, corrosion and
different CTE arise (Stavrov and Bersee ).
Another critical issue is the electrical connection essential for sufcient current ow.
Different approaches are reported, amongst others ) direct clamping, either on prepreg
or bare bres, ) bare bres coated with silver-epoxy ller or ) dipping prepreg in a liq-
uid metal bath or ) low melting point alloy pressed on bare bres (Costa et al. ;
4:
3 Overview on Fusion Joining Processes
Stavrov and Bersee ). Nevertheless, Ageorges et al. (a) concluded in the light
of cost and time for prepartion, method of choice is direct clamping on bare bres hence
clamping pressure gets into focus. They further found a direct relation between clamp-
ing pressure and resistance value which shows a minimum plateau for pressures be-
tween and  MPa for CF heating elements. Similar to that, stainless steel provides
an optimum process window around MPa (Stavrov et al. ).
For welding pressures, either a constant load or constant displacement approach are
common (Yousefpour et al. ). Where constant pressure provides full pressure con-
trol but hardly predictable variations in nal thickness, constant displacements achieve
dened geometry but uctuating pressure during welding process (Stavrov and Bersee
). For aerospace applications, especially concerning lap joints at the outer skin
with great importance of aerodynamical surface, maintaining the shape is of primarily
importance and thus constant displacement control seems more appropriate.
Don et al. () report four failure modes of resistance welded specimens (in order of
increasing strength): ) interfacial failure (separating laminate from heating element),
) cohesive failure (through heating element), ) tearing of heating element (jump of
failure path through heating element) and ) tearing of the laminate (mainly within H AZ).
Advantages of resistance welding lie in the fast and simple process which does not
require complex tooling or intensive surface preparation. As aforementioned, in terms
of maintenance and repair, rather easy and portable equipment contributes additionally
(Yousefpour 6). Furthermore, heat is mainly generated at the interface layer, there
is no restriction to at surfaces to be joined, joins found to be insufcient (aws/de-
fects) as well as damaged parts can easily be reheated and replaced/repaired or even
completely disassembled and recycled. Out of scope but possible is the application as
curing support for thermosets or adhesive bonding as well as to join hybrid/dissimilar
materials (PDL 8, 86; Hou et al. b).
Process drawbacks are potentially occurring effects, just as preferential heating which
has various forms of appearing with the same result of incomplete welding hence weak
joints. Either due to poor contact, broken heating element or reduced heat transfer at the
end of the heating element to air and thus steep thermal gradients. The latter is called
edge effect describing a rapid melting process in that area and consequently contact
of heating element and bre leading to an immediate current leakage. Melt ow prop-
agation does its bit as it “benets” from edge effects and propagates from the out- to
the inside (E. Eveno and Gillespie 88). The longer presence of melt at the outsides
rises the risk for current leakage. Moreover, matrix at the critical outer positions might
42
3 Overview on Fusion Joining Processes
reach thermal degradation temperature even before the centre is melted. Therefore,
edge effects contribute twice. Countermeasures were discussed in literature, e.g. ac-
tive cooling, insulation or clamping closely at the part to overcome the biggest difculty
in resistance welding of carbon bre reinforced composites. Another option is to com-
pletely electrically insulate the heating element via glass bre, Thermabond
®
or coating
layers (Stavrov and Bersee , ).
Although most experimental data was achieved with coupon test, main eld of ap-
plications particularly in context of this work will lie in large scale joining. Simple
extension of heating elements was found out to fail as there is no possibility of utilising
up to innite length heating elements due to Ohms law (Swartz and Swartz 8). Great-
est potential was discovered for long, narrow, sequentially welded lines/areas (Fernie et
al. ; Maguire 8). Sequential resistance welding (SRW) stepping the entire pro-
cess was introduced achieving double lap joints up to . m length and thus proving
feasibility for large-scale resistance welding application, although further improvement
is necessary especially part alignment, cost and cycle time (Taylor and Davenport
). This should also overcome the problematic nature of power/pressure require-
ments, which would have to be scaled up the same way as the specimens which is
about two orders of a magnitude (McKnight et al. ). Lambing et al.  (;
) proposed a pressure controlled automated resistance welder (ARW) with active
nitrogen cooling which simultaneously prevents oxidation of the heating element.
Other difculty for large areas is a uniform temperature distribution which was iden-
tied by Ageorges et al. (a). They proposed a criterion equation to compute maxi-
mum length of a weld line preventing thermal degradation of matrix as aforementioned.
This reads
L
max
= 2
T
max
T
min
T
1
L
!
(Eq. .)
with T
max
as degradation and T
min
as melt temperature, T
1
as temperature difference
between centre point and penetration area and L the half length of tested specimen.
The latter lead also to another problem: a continuous resistance welding is hardly
possible or at least very difcult and thus not surprisingly not very common. Not only
that heating elements have an inherited limitation in length as showed above, series--
connected heating elements to achieve a quasi-continuous process requires complex
equipment prone to disturbances, introduces many more process parameters and is
much more difcult to simulate. After Yousefpour and Octeau published a patent on
that topic in , the NRCC claimed three years later “[t]he focus of current research is
43
3 Overview on Fusion Joining Processes
on developing resistance welding, a new technology for joining large parts in a contin-
uous/progressive manner.”() However, not much scientic work and studies have
been caried out since then. Even research on skin/stringer joints were conducted out
with cut-out specimen instead of full-scale tests (Dubé et al. ).
One remarkable exception is presented by Shi et al. () whereof the PhD thesis of
Shi comprises a chapter about “Process modelling of continuous resistance welding”
(). The chain of progress got from original resistance welding process to sequential
resistance welding and nally to continuous resistance welding (CRW). Although CRW
simplies advanced SRW further, both exhibit an increased complexity for the target to
join large areas with reasonable power and pressure efforts (Shi et al. ). CRW es-
tablished by Shi () combines single-piece heating elements with a number of cop-
per blocks positioned one after another parallel to the welding direction one above
and one underneath both adherends for single lap joints maintaining a clamping dis-
tance of mm. Two copper wheels close the circuits one after another to start and
interrupt heating in the respective elements while applying clamping force/pressure of
 N (. MPa). Wheels motion along the welding line creates a continuous process
(Figure .).
(a) Schematic Welding Progress (b) Cross Section and Components
Figure . Continuous Resistance Welding (Shi )
Contact resistance, heat transfer efciency, welding voltage and welding speed were
found as inuencing parameters on temperature distribution particularly the latter two
–, which was observed quite uniform along, but very unequal transverse the bonding line
with a signicant edge effect. The size of the blocks affects the selection of welding
parameters (Shi et al. ).
Nevertheless, resistance welding technique is typically applied in automotive sector,
for plastic pipes, containers and medical devices (Ageorges et al. ). “It is believed
that the combination of sequential resistance welding and impulsive resistance welding
44
3 Overview on Fusion Joining Processes
as one welding system can provide a high-quality weld for large aerospace structures
and replace traditional techniques such as adhesive bonding and mechanical fastening.
However, this has yet to be demonstrated. (Yousefpour et al. ) Costa et al. ()
see resistance welding also as “[...] a very promising joining technique for aerospace
application”, but in the light of a continuous joining this method needs to catch up with
other approaches.
.. Induction Welding
Faraday rstly discovered electromagnetic induction while James Clerk Maxwell found
four differential equations describing these effects (Bayerl et al. ) and thus build-
ing the backbone of a controlled induction process. Electrically conductive materials
exhibit induced recirculating eddy currents when exposed to an high-frequency alter-
nating current hence magnetic eld. In turn, resistive effects of induced currents create
eventually the heat required to melt the matrix and form under pressure the bondline.
The exact source of heat in composites is controversial: reported is bre heating (Joule
loss) and junction heating (either dielectric hysteresis or contact resistance heating)
with different opinions concerning the major mechanism. In magnetic materials, addi-
tional hysteresis losses can occur (Ahmed et al. 6; Shridhar Yarlagadda et al. ).
With this heating method, transferable heat is increased drastically, e.g. by over three
orders of a magnitude compared to conduction (Benkowsky ).
Equipment consists of ve basic components: ) Induction generator (converting
to high-frequency output frequency ranging from -8 MHz, in general from - MHz),
) heat exchanger to carry away heat and cool, ) work coil (providing magnetic eld),
) pressure device and ) xture to hold joining adherends and ensure a good consol-
idation (Rotheiser , -8). Process characteristics are a heating stage reaching
maximum temperature, a slow but subsequent cooling due to heat convection to sur-
rounding air sets in before joining pressure is applied, e.g. by an externally cooled roller,
and nal consolidation until cooling to ambient temperature can be observed (Rudolf
et al. ).
As aforementioned, necessary prerequisite for this method is the presence of electri-
cally conductive joining adherends. Since carbon bres are electrically conductive, they
can be used as internal energy susceptors (Ageorges and Ye , ). Yet, anisotropy
and the need for closed loops appear as some difculties. Unidirectional even car-
bon bre reinforced composites cannot get welded, plain weave materials shows better
heating behaviour and eventually ±°UD plies achieved superior properties concern-
45
3 Overview on Fusion Joining Processes
ing heating just as multi-directional CF plies. What all have in common due to suscep-
torless
set-up is a heating of the entire part which in turn requires adequate tooling
to prevent adherends from deconsolidation and bre disturbance (Rudolf et al. ;
Vervlied and Heward ). For two reasons, susceptors inserted in the bondline can
be favourable: rstly, to induce preferential heating at the interface and, secondly, for
non-conductive and/or low permeability polymers. These can be deployed in form of
tapes of thermoplastic with ller particles, f.i. iron, stainless steel, ceramic, ferrite or
graphite (Schwartz ). Another approach was introduced by Leatherman () us-
ing a metal mesh providing preferential heating but also connection between two incom-
patible thermoplastics. Further studied is the so-called EMAWELD
®
bonding comprising
a thermoplastic paste with metal particles (S. Yarlagadda et al. 8; Costa et al. )
or nickel-coated graphite/J-polymer prepreg layers (Benatar and Gutowski 86). How-
ever, Hou et al. () point out modied attitudes due to metal particles which can lead
to weaker mechanical properties since these particles act as micro-cracks and notches.
In literature, several process parameters are named for induction welding: Ahmed et
al. (6) list current frequency, input power, welding pressure and time. According to
that, frequency is crucial since it is not only responsible for induced eddy currents but
also inuences penetration depth. Thus, a higher frequency causes higher power but
lower penetration this phenomenon is called skin effect” (Bayerl et al. ). Con-
trary to dielectric heating, frequencies are in any case higher to avoid that plastics are
affected and exhibit direct heating (Stokes 8).
Power denes the amount of energy applied for heating. It can be determined from
an in- or output point of view:
P =
u
2
ind
R
f
=
(2π f µH(I)A)
2
R
E = P
w
·t = m
w
·c ·T (Eq. .)
with the current dependent magnetic eld H(I), cross-section of affected zone A, per-
meability µ and electrical resistance R
f
of the joining adherends, respectively, the mass
of workpiece m
w
, its specic heat and corresponding temperature increase T (Rudolf
et al. ; Ahmed et al. 6). The higher the pressure, the better the consolidation
hence bond strength until it reaches a maximum. The residence time behaves in the
same way, three stages can be distinguished whereof one exhibits the maximum for
. Bayerl et al. () remark that the term susceptorless” introduced by Ahmed et al. (6)
refers to no additional foreign materials added for heating purpose; however, susceptors need
to be present anyway to enable an inductive heating, in this case these are already incorporated
in electrical conducting carbon bres
46
3 Overview on Fusion Joining Processes
weld strength (Rudolf et al. ; Zach et al. 8).
Bayerl et al. bring susceptor inuences into play as well as extended observations
from above that “[t]he main inuences of the induction setup originate from the coil
geometry, the applied electrical power and the coil current. In addition, the frequency
and the coupling distance play an important role. () The reduction of the latter can
cause higher temperature increases hence non-uniform heat distribution. Same effect
has an current increase. The common thread for all this phenomena is the coil geom-
etry since it is decisive for basic magnetomotive force hence induction and heat. This
parameter has been therefore subject to several studies investigating the correlation be-
tween coil geometry and heating pattern (Lin et al. ; Rudolf et al. ; Benatar and
Gutowski 86). Nonetheless, standard coil shapes have been reported and deployed
regularly (Figure .a) and “[...] the modication and improvement of coil geometry
and machine parameter adaption are essential for an increased applicability of induc-
tion heating in the composites industry. (Bayerl et al. ) Despite some constants
like copper material or integrated water cooling, new developments are reported in that
eld: Rotheiser () presents a hairpin coil for joining large at sheets (Figure .b),
frequency variation approach was introduced (Puyal et al. ) and additive manufac-
turing techniques could open the eld to complete new coil designs (Bayerl et al. ).
(a) Single-Turn, Multi-Turn and Pancake Coil
(Sanders 8)
(b) Hairpin Coil (Rotheiser , )
Figure . Various Coil Designs
Apart from coil geometry, Rudolf et al. () provide an intensive research study on
exact inuences of different parameters.
Other effects reported are proximity, ring and edge effects. All describe stimulation
of the magnetic eld, e.g. interference with close conducting elements (Rapoport and
Pleshivtseva 6, -). Another possibility to focus magnetic elds is the utilisation of
magnetic ux concentrators which increase coil efciency and limit the area affected
by the magnetic eld more strictly (Haimbaugh , 6-) .
Despite carbon bres are processed in composites, susceptor particles can be still
added for better control of the heating stage. By this means, Worrall and Wise ()
47
3 Overview on Fusion Joining Processes
introduce a novel approach for focused induction heating at a dened interface layer.
They use a stack-up of unidirectional plies separated by insulating glass bre plies ex-
cept near the interface layer, where ° and ° UD plies are placed in direct contact to
allow induction heating there. Therefore, range of application can be expanded to par ts
where risk of thermal degradation was estimated too high so far.
Rotheiser (, -6) assigned the following advantages to induction welding:
High production rates– weld times of to  seconds
Joint strength can reach very high levels
Permanence (even hermetic seals) as well as reopenable joints possible
Flexibility to similar or non-similar materials
Freedom in adding ller materials
Shape freedom– complex contours or even hidden joints are possible; at least the
coil must be able to pass the bond line
Clean atmosphere– no ventilation/solvent removal required
Process freedom– thermoplastic parts of all manufacturing processes can be
inductively joined
Large part capability and continuous processes are deployed
Non-destructive testing of produced joints is easily achieved
Loose tolerances due to owing melt material
Very precise process control
Especially the latter opens the eld for a high potential of automation and closed-loop
applications (Rodgers and Mallon ). Several utilisations of automated systems
and robots in combination with induction joining processes were reported (Moser et
al. 8; Wijngaarden ; Bayerl et al. ; Rudolf et al. , ).
Ahmed et al. (6) emphasised the advantage of this non-contact welding process
and the opportunity to clearly dene heat affected zones. Yousefpour et al. () point
out the fast and clean technique and the ability to reopen the joint in case of inaccu-
rate welds, internal repair or part replacements. Particularly exibility and versatility is
a major point: induction welding can provide structural, pressure-tight joints for small
or large parts, three-dimensional and complex bond line contours and geometries, ei-
ther as spot or continuous welding technique resulting in high quality stress-free welds.
Performance can be increased by adding resin/ller material and enhance preferential
heating and allow loose tolerances and various manufacturing methods. Not only weld-
ing times are short, associated costs for machine/worker stay therefore low, too, as
48
3 Overview on Fusion Joining Processes
well as reject rate and energy consumption (PDL 8, ). Border and Salas (8)
report high lap shear strength for APC- composite and Rodgers and Mallon () as-
sign good repair attitudes to inductive joining with achieved  to 8 % compressive
strength in impact damaged parts under vacuum bag conguration. Lewis () ac-
cented portability and applicability in eld repair.
Literature often adduces main drawbacks as inserted susceptor material with its high
(recurring) costs and possible (negative) inuence on joint properties (Rotheiser ,
6; PDL 8, ; Yousefpour et al. ; Stokes 8). For carbon bre compos-
ites and no envisaged enhancement or preferential heating behaviour with auxiliary ller
material, no additional insert materials are required. In such a case, a rst major issue
arises: if necessary ller materials are deployed, foreign particles are introduced in the
joint region. This is seen highly critical by certication authorities since possible initial
points for cracks or stress concentrations are introduced.
Another point is the bulk heating nature causing signicant volumetric heating unless
precautions are take, e.g. with novel focused heating approach by Worrall and Wise
(). Especially concerning aerodynamic surfaces but also global tolerances, geom-
etry changes are not desired. Even points above matrix decomposition temperature
can be reached near the induction coil, while inner plies merely reach melting tempera-
ture due to the weaker magnetic eld (Matsen and Hodges 8). Avoiding overheating
Duhovic et al. () introduced an air cooling with compressed air through the induc-
tion coil to keep the surface (exposed to strongest magnetic eld) under decomposition
temperature.
Thirdly, experiments conducted by Airbus Helicopters using an induction welding pro-
cess were reported quite insufcient since the metal mesh deployed as lighting protec-
tion caused preferential heating on outer surface of the aircraft skin (6). To over-
come this problem, deployment of lightning protection must be postponed after induc-
tion joining step challenging well-established scheduling and planning of lay-up and
manufacturing of the aircraft skin.
Inductions versatile nature is present in other applications apart from TPC welding:
thermoset curing, selective heating, triggering effects in polymers and inductive mould
heating are among common utilisations (Bayerl et al. ). The latter leads to the fourth
major issue: tools used for tape laying of UD plies are usually made out of metal. Preva-
lently, the iron-nickel alloy “Invar” is preferred due to its very low CTE similar to carbon
bre composites and concurrently comparably high thermal conductivity λimportant for
fast and efcient curing processes.
49
3 Overview on Fusion Joining Processes
Table . shows properties for electri-
cal and thermal conductivity (ρ / λ), re-
spectively, and coefcients of thermal ex-
pansion for Invar6 and Fused Castable
 Silica reported in literature. Both
have a similar CTE, but resistivity is much
higher for the alike insulating behaviour
of ceramics and thus no induction sets
in. Moreover, thermal conductivity of In-
var6 is one order of magnitude higher
compared to regarded ceramic material.
Table . Comparative Invar6 and Ceramic
Properties
Invar6* Ceramic
λ W/m·K  .6
ρ 
-8
m .8 ↑↑
CTE 
-6
/K < .8
Source: *Martienssen and Warlimont 6, 8;
Fused Castable  Silica: Hussey and Wilson ,

Although most far away from the induction coil, the metal tooling is directly exposed
to redeemed magnetic eld and still acts at least as massive heat sink causing much
higher energy consumption (Matsen and Hodges 8) and the still present risk of too
intensive heating of the tooling which could in turn melt/damage the outside of the part.
An overcome for this problem is either a cooled tooling or induction-unsusceptible tool-
ing materials such as ceramics (Matsen and McCarville 8), both immensely increas-
ing costs and complexity. Most of the experiments or in-eld applications are conducted
with clamped/xed but levitating parts where the mould heating is absent. In their exper-
imental set-up, Pappadà et al.  () welded the lower adherend onto a composite
base plate maintaining an induction situation where all materials are equally excited but
the distance from the coil decides how strong the induction heating appears. Mitschang
et al. () emphasised the use of a “non-conductive base plate.
Grumman Aircraft Laboraties assessed induction heating as highly suitable in aircraft
construction and repair particularly the F-A horizontal stabiliser leading edge out of
carbon bre reinforced thermoplastics possessing comparable or enhanced proper-
ties than structural parts processed with autoclave co-consolidation (Mahon et al. ;
Kagan and Nichols ). Costa et al. () conrmed suitability of induction heating
while achieving acceptable joint strength.
Rotheiser (, 6-) identies further disadvantages in the need for coil acces-
sibility hence shape limitations, possible high costs for acquisition and development of
achieving optimal process conguration as well as simultaneous heating of adjacent
metal inserts. Other cons are related to metallic ller materials like additional insertion
operations, stress concentrations and decomposition at the metal-polymer interface
which are not present in case of susceptorless induction welding.
50
Evaluation
The following evaluation of ultrasonic, resistance and induction welding assesses the
respective process attitudes mentioned above according to stated evaluation criteria at
the beginning of this work. The content refers to the same literature as aforementioned
as far as not named explicitly.
For a quantitative assessment, points are assigned to the three regarded processes
(displayed in squared brackets at the end of each criteria elucidation in the order: ultra-
sonic, resistance and induction welding). Starting with three points for the technique
with the best properties going down to two and one point for the following methods.
Similar results are regarded as equal points. Primary criteria are factored doubled un-
like secondar y criteria. If one criteria is not sufciently fullled by a process, zero points
are assigned. [US–RW–IW]
. Process Capability
.. Parameter
The following elucidation of process parameters is based on the work of Villegas et
al. () who regarded CF/PPS composites. Table . gives a comparison of PPS and
PEEK material properties suggesting a reasonable transferability of PPS results to PEEK
although authors emphasised the dependency of results on the substrates (type, thick-
ness, quality, weave) and thus difcult generalisations out of them. The intention is
nevertheless to reveal at least the magnitudes dealt with in such processes.
Table . Mechanical and Thermal Properties of PPS and PEEK
Structure ρ E R
m
ε T
g
*
T
s
*
HDT
*
[g/cm
] [GPa] [MPa] [%] [°C] [°C] [°C]
PEEK semi-cr. .
. - .8  -   -   -   
PPS semi-cr. .6
. - . 6 - 8 -  8 -   -   - 6
*
T
g
: glass transition temperature, T
s
: solidus/melting temperature, HDT: Heat Deection Test
Source: Schürmann , -;
Kaiser , 8;
Domininghaus , 
There are results for CF-PEEK composites using the considered joining methods, but
these were either not available for the author of this work (Beevers ) or only inves-
4 Evaluation
tigating one or two joining methods separately. Since material and process parameters
play a key role in achieving good welding quality, a comparative summary of different
papers is therefore doubtful. The investigated optimum process parameter by Villegas
et al. () are presented in Table ..
Table . Process Parameters for Ultrasonic, Induction and Resistance Welding
Method Cycle Time Welding Pressure Power Energy Consumption
[s] [MPa] [W] [kJ]
Ultrasonic . 8 .
Resistance  .  .8
Induction 8 .8  
Source: Villegas et al. 
Regarding cycle time, ultrasonic welding exhibits a by factor  shorter process period
compared to resistance and induction welding, respectively. Not only the heating phase
(. s) is by far the fastest compared to the others, moreover, heating primarily oc-
curs direct at the bonding interface comparable to resistance welding (no bulk heating
as for induction heating). Yet, during cooling stage, high thermally conductive metal-
lic sonotrode enhances energy dissipations. Though, Villegas et al. () point out
multiplication of cycle times for continuous ultrasonic or induction welding approaches
as actuators must be moved along the bondline. Notwithstanding, resistance welding
gets into same trouble since potential joint length considered are far beyond electrical
or power limitations. Another point to be considered is the dependency of heating time
on the weave as investigated by Rudolf et al. () and according to that, UD plies
assigned to later application exhibit longer heating times by almost one order of mag-
nitude compared to plain weave fabrics. [––]
Welding pressure and energy consumption will be discussed in sections Automation”
and Environmental Aspects”, respectively.
.. Automation
Closed Loop Capability. All three methods show good automation potential since all ex-
hibit distinctive process parameters recorded for process observation, control and mod-
ication, respectively. Resistance and induction welding uses power records unlike ul-
trasonic welding with additionally displacement data and part’s mechanical impedance
52
4 Evaluation
to determine joint quality. Induction welding offers a certain distance margin since it is a
non-contact methods whereas ultrasonic’s sonotrode needs a certain contact pressure
for achieving high quality weld. For more complex shaped parts with rough tolerances,
utilisation of distance sensor are conceivable. Main drawback against this background
is contacting of heating element in resistance welding which requires high precision po-
sitioning. [––]
Robotic Capability. Basically, all systems are realisable with robots. Endeffector size
and weight range within technical limits of standard robots as well as process forces.
DLR demonstrated on the JEC World  exhibition a fully automated robot cell join-
ing thermoplastic composites with resistance welding (DLR b). Automated induc-
tion welding robots have been reported by Moser et al. (8), Wijngaarden () and
Bayerl et al. (). Automated ultrasonic applications are reported by Gardiner (),
Offringa () and AM ().
One important factor will be welding pressures since robots are not capable of high
process forces. Results presented in Table . show marked differences: resistance
welding needed low pressures due to direct interface melting and low desired squeeze
out followed by induction welding. For both, pressure applied is merely for consolidation
and preventing delamination. In contrast, pressure in ultrasonic welding serves both,
sufcient contact for good energy transmission as well as ow and fuse of molten en-
ergy directors with the adherends (Avraham Benatar et al. 8) and thus directly related
to the process’ heating. Villegas et al. () determined optimum heating and consol-
idation pressures for CF/PPS composites as . and . MPa, respectively. Computing
process forces for a use case according to H pressure and sonotrode area yields
about  N. Compared to realised robotic friction stir welding machines with high pay-
load industrial robots and pressures up to  kN (Völlner , 6; Zaeh and Voellner
), doubts can be redeemed but should still kept in mind. Also concepts with force
control devices have been developed (C. B. Smith ). [––]
.. Process Chain Adoption.
Tooling. For ultrasonic and resistance welding, standard metal toolings can still be used
unlike inductive technique. “The dies should not be susceptible to inductive heating so
mat heating is localized in die retort. We prefer a ceramic that has a low coefcient
of thermal expansion, good thermal shock resistance, and relatively high compression
53
4 Evaluation
strength, such as a castable fused silica ceramic. (Matsen et al. 6) Several patents
introduce ceramics as tooling material for inductive heating processes (Matsen and
McCarville 8; Matsen et al. 6; Matsen , ; Matsen and Hodges 8) in-
creasing complexity and costs. Since predominantly metal forms have been utilised so
far, complete new toolings need to be produced making a simple, fast and cheap imple-
mentation in series production impossible. [––]
Surface Preparation. Accurate cleaning of joining surfaces is enough for ultrasonic and
induction welding. Nature of of “dissimilar” joining when incorporating a metal mesh in
a composite matrix rises the need for good adhesion between both materials. This zone
is already regarded as the most critical for failure initiation particularly fatigue which is
more severe in aerospace applications (Dubé et al. 8a, 8b, ). The impor-
tance and severity of surface preparation for good metal-matrix adhesion in resistance
welding is emphasised repeatedly in literature (Hou et al. a; Delgado Labrandero
, ; Freist , -) and therefore requires a costlier additional manufacturing
step. [––]
Manufacturing. Ultrasonic and resistance welding do not require re-scheduling of man-
ufacturing process steps whereas lightning-strike protection in the top layers of the skin
lay-up turned out to be a severe issue when regarding inductive welding. Airbus Heli-
copters (6) conducted such experiments coming to the conclusion of preferential
heating rather on the outside of the skin and not at the interface. In addition, maintain-
ing outer aerodynamic shape becomes just as critical as achieving joining temperature
at the bonding interface. Consequently, lightning-strike protection deployment must be
outsourced during skin lay-up face and raised at a later stage of production. Again, a
simple, fast and cheap implementation in series production is hardly possible. [––]
. Aircraft Applicability
.. Geometry
Large Scale Continuity. Ultrasonic and induction welding are capable of large geome-
tries via a moving sonotrode or coil, respectively, and such applications have already
been reported, f.i. with subsequent consolidation rollers and sensor implementation in
complete endeffectors. Resistance welding is limited in welding length due to factors
54
4 Evaluation
like Ohm’s law and leakage current tendency. Continuous approaches have been re-
ported very rarely and show more complex experimental set-ups compared to the other
two techniques. SRW and ARW are known in this context whereof the rst one produced
joints with a maximum length of . m (Taylor and Davenport ; Lambing et al. )
which is comparably short to aircraft dimensions. [––]
Lap Joint Design/Accessibility. Envisaged are lap joint designs of which all three bond-
ing techniques are able to produce. The point of accessibility is regarded as the addi-
tional space needed despite the actual bonding surface: the sonotrode is just as big as
the desired joint width is and thus no extra space is needed. Largest ultrasonic welders
are reported as about . m × . m (Rotheiser , 8) which should be sufciently
big to achieve one-shot welding in width direction. What must be kept in mind is the cor-
relation of parts curvature and size of the sonotrode: greater curvatures require shorter
sonotrodes whereas little curvature enables longer sonotrodes at least a planar con-
tact surface for optimum transmission of ultrasonic vibrations must be guaranteed. De-
pending on edge effects due to eddy currents induction, needed space is the joint width
or slightly broader, too. More crucial is the coil design, number of turns and supply pipes
for water cooling which can extent the required space signicantly also in length. Al-
though the heating element is entirely covered by adherends, contacting requires extra
space at least on one side. In case of the regarded continuous approach with copper
blocks as connecting element, additional space needed is signicant and required ac-
cessibility from two sides is hardly manageable with the existing tooling concept. Util-
isations of consolidation rollers for achieving welding pressure can be seen as similar
for all approaches. [––]
Tolerance Management. According to Rotheiser, induction welding is well suited for
closing gaps and voids of irregular surfaces allowing loose tolerances of pre-manu-
factured parts, yet, high surface qualities of nal parts “... are particularly difcult to
achieve with processes that have only one controlled surface [...] These are also meth-
ods in which the type of tolerance needed for such joints is held only with difculty.
(, 6, 6) Therefore, bulk heating leads to uncertainties in nal shape especially
on upper and side surfaces although a consolidation pressure is applied. Comparably
local heating of resistance and ultrasonic welding causes usually only little volumet-
ric heating near the interface hence maintaining easily basic part’s shape. Considering
compensation of loose tolerances, although regarded critically from a process point of
55
4 Evaluation
view, ultrasonic welding performs best as the inserted resin lm as energy director pro-
vides additional matrix material to ll porosities. [––]
.. Performance
Lap Shear Strength. The static capabil-
ity is often quantied by the lap shear
strength (LSS). Villegas et al. () in-
vestigated LSS for identical CF-PPS com-
posites with ultrasonic, resistive and in-
ductive welded specimens as indicated
in Figure .. Although desired matrix
system shall be PEEK, this work is of in-
terest since it directly compares same
substrates with different joining methods.
Thus, ultrasonic and induction welding
achieved similar strength values about
. MPa whereas resistance welding
dropped down to . MPa with consid-
erable higher scatter than the other two
methods (see also Certication).
Ultrasonic Resistance Induction
0
5
10
15
20
25
30
LSS [MPa]
Figure . Comparative LSS of Ultrasonic,
Resistive and Inductive Welded
CF/PPS Specimen (Data: Ville-
gas et al. )
Dubé (, ), Ageorges and Ye
(, -) and Yousefpour et al. ()
collected LSS values for APC- (CF/PEEK)
composites and different joining meth-
ods. Computed average values are pre-
sented in Figure . resembling the gen-
eral trend from above presented results
for CF-PPS composites. In both cases, ul-
trasonic and inductive welded specimens
possess superior properties compared to
resistance welding.
Ultrasonic Resistance Induction
0
10
20
30
40
50
60
70
80
LSS [MPa]
Figure . Averaged LSS values for APC-
(CF/PEEK) Specimen
Scatter is noticeable higher than for the comparative study of CF/PPS which can be
accounted for different materials and set-ups in various studies, averaged values are
computed out of. This is another argument for the reasonability of attaching values to
the study of Villegas et al. ().
56
4 Evaluation
Undoubtedly, resistance welding behaves inferior again resulting from poorly welded
areas on the outer edges due to edge effects accounting for about - % reduction in
joint area hence LSS (Ageorges et al. b). This ts quite well with obtained average
values. With respect to computed standard deviation, LSS for ultrasonic and induction
welding can be seen as equal again, albeit with slight advantages for the ultrasonic
approach. All in all, the conclusion of consistently higher static strength values for ul-
trasonic and inductive welding can be drawn with a distant resistant method. [––]
DCB. Values of interlaminar fracture toughnesses (G
Ic
) are presented in Table .. De-
spite intensive research, no comparative study of all three joining types could be found.
Therefore, presented results must be treated carefully since differences in material and
congurations might falsify comparability. No value for inductive welded CF/TP spec-
imen has been found either. Since no conclusion can thus be drawn, no points will be
awarded.
However, Harras et al. (6) achieved
highest G
Ic
values with optimum ul-
trasonic welding parameters up to
. KJ/m
. Also Jakobsen et al. (8)
achieved slightly higher fracture tough-
ness than reference CF/epoxy system
investigated by Markatos et al. ().
Already from this rough consideration
(Table .), it is obvious that there is
no signicant reduction of mechanical
properties; quite the contrary.
Table . Interlaminar Fracture Toughness
G
Ic
for Different Joining/Material
Congurations
Laminate Joining Method
G
Ic
[kJ/m
]
CF/Epoxy
(Baseline)
Autoclave
Co-consolidation
.*
CF/PEEK Resistance W. .
CF/PEEK Ultrasonic W. .
Sources: *Markatos et al. ;
Jakobsen
et al. 8;
Harras et al. 6
Fatigue. Again, the rst regarded study
(Villegas et al. ) provides a starting
point and magnitude (Figure .). All three
welding techniques lead to more or less
similar fatigue behaviours. Ultrasonic ex-
hibits a slightly deeper decrease in % LSS
compared to the other two. Eventually, an
endurance limit was determined at about
 % LSS and run out samples (
6
cycles)
where tested statically with no signicant
fatigue damage found.
10
4
10
5
10
6
40
45
50
55
60
65
70
No. of Cycles to Failure
S
max
[% LSS]
US
RW
IW
Figure . S-N Curves of Differently Welded
CF/PPS Specimen (Data: Ville-
gas et al. )
57
4 Evaluation
Similar behaviours where obtained for CF/PEEK specimens with endurance limits be-
tween and  and  % (Yousefpour and Hojjati ). The similarities of S-N curves in
general for different welding technologies is proven by Withworth (8) Villegas et al.
(). Therefore all welding types possess a comparable fatigue behaviour with only
marginal differences. [––]
Failure Modes. Before nal coupon fail-
ure occurs with highest strength (inde-
pendent of welding quality), other failure
modes appear (Figure .). Interlaminar
failure represents a failure within the lam-
inate, the heating element or both. Lower
strength is observed for interfacial fail-
ures between adherends and heating el-
ements depicting an imperfect bonding
(Stavrov and Bersee ).
Figure . Failure Modes in Lap Shear Tests
(Meng Hou and Friedrich )
These failure modes are applicable and reported for all three welding methods (Strong
et al. ; Don et al. ). O’Shaughnessey et al. (6) showed occurrence of only in-
terlaminar failure modes for all three types of welding techniques when recommended
process parameters are applied. If interfacial failure due to poor adhesion between
metal mesh and matrix can be excluded with sufcient pre-treatment, all process pro-
vide favourable results.
.. Certication
Reproducibility/Scatter. Number of parameters as variables and its uctuation plays
an important role for reproducibility. Ultrasonic welding inherits variables in frequency,
amplitude, energy director shape and height/thickness, welding pressure/force and vi-
bration/consolidation time (Troughton 8, ; Villegas and Palardy ). Induction
welding incorporates frequency, generator power, distance between coil and laminate,
induction coil geometry, number of coil turns, coil position and ux concentrator, lami-
nate structure and material, welding pressure/force and time, cooling rate which is rep-
resented by ux of compressed air for surface cooling and ux of water through the coil
(prevent overheating of coil and laminate by convection) and consolidation rollers (cool-
ing time to solidication) as well as deployment of additional susceptor particles (Rudolf
et al. ; Ahmed et al. 6; Troughton 8, ). Resistance welding shows param-
eters in input power (current/voltage), resistance (length/diameter/material), clamp-
58
4 Evaluation
ing/welding pressure (Stavrov and Bersee ). Despite not explicitly mentioned, just
as for adhesive bonding, surface preparation is crucial to achieve good metal-matrix ad-
hesion. In the same manner, its rather unpredictable behaviour dramatically inuences
the joint’s mechanical properties just as occurrence and position of poorly welded ar-
eas (Ageorges et al. b). In the comparative study of the three methods (albeit for
CF/PPS composites) by Villegas et al. (), a similar conclusion to previous eluci-
dations can be drawn regarding lap shear strength as shown above: not only LSS of
resistance welding parts is considerable lower compared to the other two, but scatter
shows a much higher value. In the test series, ultrasonic process with its less parame-
ters was operated near the optimum with very low scatter. Slightly higher but still low
scatter was obtained with induction welding and its more numerous but good control-
lable variables (Figure .). When regarding continuous joining, the feed velocity is an
additional variable for all. [––]
Online Inspection. Most recent developments in resistance welding at DLR Augsburg
(a) go in the direction of monitoring power respectively voltage/current data to get
an insight of weld quality and its improvement. Lambing et al. (), Holmes et al.
() and Tackitt and Gillespie (6) introduced a non-contact monitoring of the soft-
ening process via ultrasonic probes. Similar approaches with power (current/voltage)
are reported for induction welding (Ahmed et al. 6) expanded by pyrometers directly
mounted near the induction coil (Bayerl et al. ; Moser et al. 8) or making re-
course of the impedance behaviour of the entire system (Puyal et al. ). “Most ultra-
sonic welding machines nowadays feature fully programmable, microprocessor control
to program and monitor all welding parameters. Some machines monitor to adjust the
entire process every millisecond. (Troughton 8, ) Beyond that, dynamic mechan-
ical impedance gives indication of molten polymer ow (Benatar and Gutowski 8).
Villegas () introduced an in-situ monitoring method assessing power and displace-
ment data recorded by the microprocessor-controlled welder for quality inspection. In-
sofar, the advantage of ultrasonic machines is the already implemented microprocessor
system whereas custom-made extensions need to be utilised for the others. However,
all three show enough possibilities for a sufcient online inspection. [––]
Foreign Object Issue. Only induction joining works theoretically without any additional
ller materials regarding carbon bre composites. If a complete susceptorless induc-
tion welding is possible in the current case needs further investigation. If not, metal-
59
4 Evaluation
lic particles/meshes are needed in the bondline to induce preferential heating. These
are seen as possible initials for micro notches and cracks and thus undesired from
certication authority’s perspective holding true for resistance welding as well with
its heating element. Dubé et al. () investigated fatigue behaviour of resistance
welded carbon composites and found delaminations always located at the weld inter-
face. Despite observed good adhesion between TiO
coated metal meshs and the poly-
mer, coating tended to separate from the metal mesh base material. For stainless steel
meshs, even poor adhesion was found just as striations suggesting crack propagation
in through-thickness-direction and peel stresses causing debonding of the heating ele-
ment at the edges (8). Arising issues are galvanic corrosion, different mechanical
properties as well as induced stresses due to different CTEs. Unlike ultrasonic welding
requiring energy directors of additional matrix material, no “foreign material is intro-
duced. Only in case of too thick matrix lms, resin rich regions can reduce mechanical
properties. Yet, this can be overcome by optimisation. Similar to previous elucidations,
additional weight penalty must not be neglected for higher density ller materials.
I [––]
. Secondary Criteria
.. Investment
Acquisition/Equipment Complexity. Ultrasonic welder only need an electrical power
supply for generating high-frequency voltage analogical to resistance welding. However,
contacting for RW is more difcult and crucial for welding quality. Since both methods
generate heat predominantly at the interface and desirably no bulk heating sets in, the
consolidation unit only needs to apply a welding pressure and no additional cooling is
necessary. Induction welding not only requires electrical power but also water cooling
and ideally compressed air for surface cooling to prevent adherend surface overheating
(Duhovic et al. ), since bulk heating of the part set in, consolidation rollers need to
be water-cooled for rapid solidication and maintaining the desired basic shape of the
components. Thus, this procedure has by far the most complex equipment requirement.
I [––]
. necessar y due to strong heating of copper; so-called cold inductors with low electrical con-
ductivity like iron exhibit less heating but higher electrical resistance hence power losses
60
4 Evaluation
Recurring Costs. Despite expenses for electrical power (see Parameter”), additional
resources are needed consistently. Ultrasonic technique requires energy directors, e.g.
in form of a thin PEEK lm, resistance welding requires heating elements and in case of
preferential heating with susceptors, e.g. metallic particles are necessary for inductive
welding. For the latter, the availability of compressed air and cooling water brings addi-
tional costs. [––]
.. Fibre-Fairness
All three processes create no bre interruptions in contrast to mechanical fastening
methods (basis for potential improvements). Ultrasonic welded parts can exhibit im-
prints of the sonotrode due to the applied pressure (Fischer et al. ). However, this
phenomenon is rather an issue for spot welding and not for a continuous process with
moving sonotrode along joint line. In such a case, potential change in thickness would
occur over the whole length and local effects can be excluded. On the other hand for
resistance and induction welding, possible ller materials could introduces interference
effects due to dissimilar materials between carbon bre and metal meshs/particles.
.. Heating Characteristics
Heat Affected Zone (HAZ). The heat affective zone for all three methods can be easily
characterised either by size of the sonotrode, heating element or coil. However, there
are distinctive differences in niteness. The sonotrode directly applies vibrations which
are converted into heat. Through the sharp limitation of the sonotrode geometry and
the deployment of ED material, the heat affected can be limited to a distinctive area.
6:
4 Evaluation
Unlike resistance and induction weld-
ing show either poorly welded areas
(see Performance”) or additionally induc-
tively inuenced areas by diffusion of the
magnetic eld. The latter can be corrected
but not eliminated by magnetic ux con-
centrators. Nonetheless, such an equip-
ment increases weight and costs of the
endeffector. [––]
Figure . Magnetic Flux Concentrator
(Ahmed et al. 6)
Heating Curve. Levy et al. () simulated heating behaviour during ultrasonic weld-
ing of PEEK composites with triangular energy directors showing a clear initialisation
and concentration of heating at the EDs, even in case of assumed equal stiffnesses of
composites and neat resin EDs. Furthermore, a quadratic inuence of the vibration am-
plitude on the heating rate was proven. The data revealed holding force as instrument to
adjust the maximum temperature in the EDs especially since temperature is approx-
imated asymptotical without overshooting (Figure .6a) and lower temperatures for
atter ED angles. This leads to the assumption that a at ED shows an even lower and
thus less dangerous heating behaviour in terms of thermal degradation by overshooting
(Figure .6b). Khmelev et al. () conrms the inability of a welding process under
too high static pressure leading to an even more enhanced damping of the oscillatory
system hence decrease in vibration amplitude respectively input energy necessary for
the melting process.
(a) Maximum ED Temperature over
Time for Different Welding Forces
(b) Maximum ED Tip Temperature over
Time for Different Angles of EDs
Figure .6 ED Temperature Development during Ultrasonic Welding for Different Conguta-
tions (Levy et al. b)
Basically, the mechanism can be seen as self-stabilising: EDs act as heat initiators
62
4 Evaluation
disappearing when exposed to ultrasonic vibrations and subsequent melting which in
turn causes a diminished heat generation and a quasi-constant temperature at melting
point level for a certain period of time until heat conduction/consolidation sets in.
Though, particularly semi-crystalline materials show a sharp melting point due to ad-
ditional energy for breaking up crystalline structure (see Section ..). Vice versa, solid-
ication appears very abruptly due to sudden recrystallization of molecules. Moreover,
their orderly molecular structure absorbs vibrational energy unlike amorphous plastics
with lower attenuation (Dukane , ).
Rudolf et al. () investigated the in-
duction heating and determined the four
stages of heating (Figure .) whereby
the constraints for temperature points are
given by:
T
m
< θ
1
< T
d
θ
4
< T
cry
T
m
< θ
2
Since there is no asymptotical conver-
gence rather than a peak in context
with the high heating rates due to the by
far larger amount of transferred heat –,
achieving the appropriate process enve-
lope with the requirement on low cycle
times is more difcult.
Figure . Typical Temperature-Time-Curve
of the Continuous Induction
Welding (Rudolf et al. )
Quite a similar behaviour can be ob-
served for resistance welding. In their
investigation on continuous resistance
welding, Shi et al. () modelled and
simulated the heat generation exhibiting
a similarity in high heating rate, peaking
temperature and subsequent consolida-
tion (Figure .8) evoking same difcul-
ties of overshooting and overheating and
the higher sensitivity to disturbances. In
this case, the peak is even more distinct
due to the sudden switch off of the elec-
trical power source.
Figure .8 Temperature Development over
Time for Different Welding
Speeds (Shi et al. )
Finally, the conclusion can be drawn that ultrasonic heating behaviour provides the
63
4 Evaluation
most desirable attitude with its asymptotical convergence rather than steep increase
and peaking temperatures for induction or resistance welding involving danger of local
overheating and decomposition hence weakened joint strength. [––]
.. Maintenance
Portability plays a role as well as the ability to reopen and replace damaged parts/struc-
ture. Concerning both, ultrasonic welding has to wait in line. Once energy directors are
consumed during initial joining, no later heating with ultrasonic vibration is possible.
In addition, Lewis characterised equipment “[...] too heavy for practical in-eld work”
(Lewis ), in contrast to the other two methods regarded as easily portable (Yousef-
pour 6; Lewis ). Furthermore, possibility of joint reopening joint is given albeit
only with bulk heating of the complete part for (susceptorless) inductive welding. Re-
sistance welding therefore exhibits more practicable detachability properties using the
still existing heating element. [––]
.. Environmental Aspects
Energy Consumption. Electrical power records presented in Table . exhibit big differ-
ences. Although it must be noted that power required for resistance welding is directly
dependent on the resistance of the heating element and therewith on the length, cross
section and material, (Eq. Eq. .), and can vary therefore, the power needed for ultra-
sonic welding is still higher by factor and  compared to induction and resistance
welding, respectively. However, when determining the total energy consumed by the
process, the very low cycle time of ultrasonic benets in a magnitude that it possess
clearly the lowest value of all three techniques, followed by resistance and the distant
induction approach. [––]
Resources. As already mentioned in the section Investment”, equipment complexity
for induction heating is mainly due to the necessity of air and water cooling. Therefore,
the use of resources (electrical power, water, compressed air) is more crucial for the
inductive approach than for the other two.
Recycleability. Regarding the recycleability, no hazardous materials are introduced dur-
ing joining processes. It can be basically broken down into carbon bres, thermoplastic
64
4 Evaluation
resins and metallic meshes (stainless steel) all enable the achievment of a closed
loop recycling system.
. Resumée and Final Remarks
The evaluation matrix yield the ultrasonic welding as method of choice followed by in-
duction and resistance welding which thereby conrms the overall impression of the
assessment.
Resistance welding showed a promising approach for static welding of thermoplastic
composites, is however hardly compatible with a continuous process. Issues in upscal-
ing (size of heating element/power requirements), continuity and access concerning
this work limit this process to an unfavourable degree as well as the generally steep
heating behaviour with distinct temperature peaks and a process sensitivity to distur-
bances leading to difcult control. Research is generally progressing, though in the eld
of continuous applications it seems to have stalled.
Induction and ultrasonic welding both exhibit advantages in heating time, upscaling,
continuity, energy consumption, areal joining and access all beneting the aim of this
research.
Disadvantages of ultrasonic welding appear as need for presence of an energy direc-
tor, consequently no ability to reopen after initial joining and higher process forces are
present. However, those can be handled.
Drawbacks of induction welding turn out to be much more severe. The perk of a non--
contact process in turn does not provide a dened geometry. Heating with envisaged
unidirectional plies is reported as difcult and ineffective (weak forming of eddy current
circuits) bringing up the need for metallic insert materials (paste, particles) for prefer-
ential heating. Foreign object issues are seen more than critical by certication author-
ities. Moreover, the prior applied lightning protection consisting of a metal mesh on
the outside of the thermoplastic shells causes preferential heating there, too, as well as
the metallic tooling. Production schedule and process auxiliaries/tooling would evoke
a costly re-design and delays series production implementation. In addition, the steep
heating behaviour and need for air/water cooling are further complexing the equipment.
The tendency towards ultrasonic welding is enhanced by the current very intensive
research and high number of publications on this topic in the last few years and on
joining of thermoplastic structures for aviation applications in particular. In contrast,
65
4 Evaluation
research on induction welding is rather outmoded and large progress in this eld has
not been identiable.
Eventually, it stands to reason to persevere the ultrasonic welding approach as it com-
bines desirable benets with acceptable and solvable disadvantage whilst offering best
performance in the numerical assessment.
66
4 Evaluation
Table . Evaluation Summary
US RW IW
Process Capability  6 6
Parameters
Cycle Time
Pressure
Energy Consumption
Automation
Closed Loop Capability
Robotic Capability
Process Chain Adoption
Tooling
Surface Preparation
Manufacturing
Aircraft Applicability   8
Geometry
Large Scale Continuity
Lap Joint Design/Accessibility
Tolerance Management
Performance
Lap Shear Strength
DCB - - -
Fatigue Behaviour
Failure Modes/Detectability - - -
Certication
Reproducibility/Scatter
Online Inspection
Foreign Objects Issue
Secondary 6
Investment
Equipment Complexity
RC costs
Fibre Fairness - - -
Heat Affected Zone
Maintenance
Environmental Aspects
Total  6 
67
Pre-Testing Campaign
. First Series
First pre-testing was performed at DLR Augsburg with a mobile test stand for ultrasonic
welding (Figure .a) introduced by Dorsch ().
(a) Mobile Test Stand
(b) Specimen/Clamping Set-Up
(c) Spot-Welded Specimens: soot-blackened
 µm,  J
Figure . First Pre-Testing Campaign Set-Up
Test specimen of about 6 x  mm were clamped with two vices onto a steel plate;
in-between a PEEK lm with  µm/ µm thickness (Figure .b). BRANSON  kHz
generator of type LPe :.:T with a circular bellied sonotrode ( /8") was centred
and positioned onto upper adherend surface. Additional weights on the construction
created a contact force F
C
 N (Dorsch ). Input energies were varied between
 and  J per specimen. A welded joint could be achieved for all of them (Fig-
ure .c) withstanding shear loads applied by hand. Although no further investigations
of strength and/or formation of joints was conducted, feasibility of ultrasonic welding
of CF/PEEK composites even with  kHz system was proven for static tests.
5 Pre-Testing Campaign
. Second Series
A second pre-testing campaign was arranged at the ultrasonic welding machine manu-
facturer BRANSON at their laboratories in Dietzenbach, Germany.
Two CF/PEEK plates with a  µm /  µm PEEK lm in-between were used for static
welding tests with a  kHz system of type X with same circular bellied sonotrode
( /8"). Initial welding tests for loose adherends at various weld times (.-. s) and
forces (- N) revealed no weldability, only slight to strong melting in the con-
tact region between top sur face and sonotrode. In a second stage, similar to the rst
campaign tests, set-up was changed with a tight xturing of probes near the sonotrode
position. Indeed, in this conguration a distinct fusion bonding was observed at the in-
terface area of adherends with subsequent nearly no imprint/melting on top surfaces.
Consulting BRANSON, welding of CF/PEEK plates with at EDs is still seen critical.
Although they experienced good welding joints with PEEK material in the past all such
joints were manufactured using three-dimensional ED shapes, e.g. triangular. Estimated
heights of desired shaped EDs for this application are .-.8 mm.
Supporting this theory, CF/PEEK was already spot-welded even for aircraft applica-
tions (Palardy and Villegas 6), however, all publications stating no difference be-
tween shaped and at EDs were predominantly conducted with composites of semi-crys-
talline CF/PPS (Senders 6, Villegas and Palardy , Palardy and Villegas , Vil-
legas et al. ) This leads to the assumption of a strong material dependence and
the tendency of less good applicability of at energy directors to CF/PEEK composites.
Detailed contemplation on that issue can be found in chapter 6 (“Heating Models, Mech-
anisms and Parameters”).
. Third Series
A third pre-testing campaign at DLR Augsburg investigated the feasibility of ultrason-
ically welded lap joints with CF/PEEK material. Set-up was identically with the rst
pre-testing series (F
C
=  N); only adherends were positioned in overlap conguration
with clamping at averted edges (Figure .). Input energy was held constant at  J.
Film thicknesses were varied with  µm,  µm and  µm (x µm loosely inserted).
All three exhibited a welded joint (Figure .) withstanding shear loads applied by hand
again. Conguration with  µm lm showed the most distinct imprint of the sonotrode
on the top surface (Figure .a) whereas there could not be found such on the other two
variants (Figure .b) implying the highest top surface temperatures hence weakening
69
5 Pre-Testing Campaign
Figure . Third Pre-Test Overlap Conguration
(a) Spot-Welded Overlap
Specimen: transparent
 µm,  J. Distinct
imprint marked
(b) Spot-Welded Overlap
Specimens: soot-black-
ened doubled  µm,
 J
(c) Spot-Welded Overlap
Specimen (Backside):
soot-blackened  µm,
 J. Molten matrix
material marked
Figure . Spot-Welded Overlap Specimen
of matrix stiffness there. For  µm lm tess, molten matrix on the lower adherend
hidden edge was observed (Figure .c) indicating a far advanced melt front at the in-
terface. Once more, no further investigations of strength and/or formation of joints was
conducted despite the feasibility of static ultrasonic welding in overlap conguration of
CF/PEEK composites was proven.
70
6 Heating Models, Mechanisms and Parameters
After proving basic feasibility, in order to understand the fusion bonding process and
how it can be manipulated towards an optimum weld strength and quality, relevant inu-
ence quantities must be determined. Therefore, the following sections focus on theory
and modelling of ultrasonic welding revealing mathematical equations and variables
representing process parameters. As a consequence, presented theory shall be con-
rmed and quantied with experimental testing in following chapters.
A fusion bonding process requires molten matrix material resulting from heating. In
ultrasonic welding, heating occurs when sonotrodes mechanical deformation work is
transferred into interfacial and intermolecular friction (Villegas ), whereof interfa-
cial friction induces initial melting at the energy directors and disappears when adhe-
sion sets in (Levy et al. a). The amount of heat transferred must therefore meet the
magnitude of the required melting energy expressed as enthalpy of fusion.
6. Enthalpy of Fusion
The enthalpy of fusion denotes the amount of energy required to transfer material from
solid to liquid state with an isobar process by overcoming intramolecular forces. Start-
ing from the specic enthalpy, energy needed for melting a certain amount of polymer
follows
h
m
=
H
m
m
=
H
m
ρ ·V
=
H
ρ ·A
C
·dz
H
m
= h
m
·ρ ·A
C
·dz (Eq. 6.)
with the specic enthalpy of fusion h
m
, density ρ, contact area A
C
and incremental height dz
of temporarily molten polymer. Specic enthalpy, density and incremental height (based
on the heat conductivity of the material) are substance-specic properties hence con-
tact area as only remaining adjustable parameter to control the required amount of en-
ergy for matrix melting in direct proportionality.
6. Deformation Work
The generator ’s electrical input energy supplies the converter which in turn creates a
vibration via a piezoelectric or magneto-restrictive actuator. This vibration excites the
sonotrode pressed against the adherend top surface by a contact force F
C
. The so-
6 Heating Models, Mechanisms and Parameters
notrodes amplitude a
S,0
is thereby a factorisation of the original converter amplitude
multiplied by booster and horn gain factors.
Important to note is that ultrasonic machine manufacturers usually refer to the ampli-
tude as peak-to-peak travel distance and not as commonly dened the height of upper
or lower sine wave.
The following derivation is based on the approach by Dorsch () to determine the
deformation work W
d
starting from of the basic work denition
W
d
=
Z
Fdx (Eq. 6.)
with the force F that acts along the path of the incremental distance dx. Assuming
the contact pressure of the sonotrode on the top surface F
C
as pre-load condition and
the sonotrodes motion oscillating around this origin, the actual contact force is time
dependent and reads
F
C,a
(t) = F
C
+ F
·sin(ωt) (Eq. 6.)
whereof F
is the alternating time-dependent part of actual contact pressure. In addition,
the actual motion of sonotrode can be described as
x
S,a
(t) =
a
S,0
2
·sin(ωt)
2x
S,a
(t)
a
S,0
= sin(ωt) (Eq. 6.)
with sonotrodes peak-to-peak amplitude a
S,0
. Inser ted in (Eq. 6.) yields
F
C,a
(t) = F
C
+ F
·
2x
S,a
(t)
a
S,0
. (Eq. 6.)
Integration of (Eq. 6.) reads
W
d
=
Z
a
S,0
2
a
S,0
2
F
C
+ F
·
2x
S,a
(t)
a
S,0
dx =
"
F
C
·x + F
·
x
2
S,a
(t)
a
S,0
#
a
S,0
2
a
S,0
2
(Eq. 6.6)
with integration limits of half the peak-to-peak amplitude in each direction. By this, the
latter term eliminates itself during insertion of limits, only remaining the rst term ex-
pressed as
W
d
= F
C
·
h
a
S,0
2
a
S,0
2
i
= F
C
·a
S,0
(Eq. 6.)
This energy is put into the adherends during rst half of the oscillation (positive sign)
72
6 Heating Models, Mechanisms and Parameters
leading to the expression for the power P
d
by deformation as
P
d
=
dW
d
dt
=
W
d
T
2
= 2F
C
·a
S,0
·
1
T
= 2F
C
·a
S,0
· f (Eq. 6.8)
with the linear dependence on applied (converter) frequency f . Since there are coupling
losses between sonotrode and adherend surface, the efciency factor η
cpl
shall be in-
troduced leading to the transferred power
˙
Q
in
= 2F
C
·a
S,0
· f ·η
cpl
(Eq. 6.)
Higher degree of deformation work can
be achieved either with increased weld/-
contact forces F
C
, amplitudes a
S,0
or gen-
erator frequencies f as well as with an im-
proved sonotrode-sample coupling.
Dukane (, ) quanties transition
efciency for sonotrode/part and part/x-
ture as - % each, depending on tool
tting. Furthermore, these interfaces are
critical since they cannot be predicted and
compensated as well as disturbances in
the converter/ booster/ sonotrode unit.
Figure 6. Energy Losses in Ultrasonic Weld-
ing Process (Dukane , )
According to that, coupling efciency from the sonotrode downwards can be calcu-
lated by (n denotes the variability from one welding to another)
η
cpl
= η
sonotr part
(n) ·η
inter f ace
·η
parttooling
(n) (Eq. 6.)
6. Interfacial Friction
Since interfacial friction heat is crucial for initial melting, modelling can refer to dry fric-
tion between two solids according to C’ L as
~
F
τ
= µ ·F
N
·
~v
|~v|
(Eq. 6.)
with constant sliding friction coefcient µ throughout the motion (unless state of melt-
ing is reached), normal reaction force F
N
and relative velocity v. With respect to the
73
6 Heating Models, Mechanisms and Parameters
applied area A, (Eq. 6.) can be re-written as
~
τ = µ ·σ
N
·
~v
|~v|
(Eq. 6.)
whereof the normal stress consists of a constant and a oscillating term recalling (Eq. 6.).
Based on this, Levy et al. () developed a model for friction dissipated power on at
energy directors as
˙
Q
f ric
(x) = α
2
h
ω
π
µ
σ
yy
(x)δ u
(x)
(Eq. 6.)
with the hammering coefcient α
h
considering contact losses between sonotrode and
adherend surface, oscillation frequency ω, friction coefcient µ, vertical stress on the
horizontal interface σ
yy
and horizontal displacement δ u
(x).
The latter can be qualitatively compared to the amplitude and considering other pro-
cess parameters, (Eq. 6.) can be reduced to
˙
Q
f ric
(x) f ·µ ·
F
N
A
·a
S,0
(Eq. 6.)
exhibiting linear dependencies of frictional heat generation on the vibration/generator
frequency f , friction coefcient µ, welding/contact force F
C
and amplitude a
S,0
as well
as an indirect proportionality to the applied area A. Jiang et al. (8) proved the indirect
proportionality of the friction coefcient µ and surface roughness R
a
of which the latter
is an adjustable process parameter. It stands to reason that shaped energy directors
increase drastically the surface roughness and increase friction heat generated.
6. Intermolecular Friction
Second stage of heating is predominated by intermolecular friction. Thus, a visco-elas-
tic model of the polymer shall be established. Unlike elastic behaviour (here σ), modied
shear stress-strain behaviour (here τ) for a viscous uid is a time-dependent approach
following
σ = E ·ε τ = η
dγ
dt
= η
˙
γ (Eq. 6.)
74
6 Heating Models, Mechanisms and Parameters
When exposed to sinusoidal oscillations, strain follows stress with a phase angle differ-
ence δ hence time dependency according to
ε(t) = ε
0
·sin (ωt) = ε
0
·e
i(ωt)
(Eq. 6.6)
σ(t) = η
˙
ε = η ·
d
dt
(ε
0
·sin (ωt)) = η ·ε
0
·ω ·cos (ωt) (Eq. 6.)
σ(t) = η ·ε
0
·ω ·sin (ωt + δ ) = σ
0
·e
i(ωt+δ )
(Eq. 6.8)
Since viscous loss occurs, the elastic modulus is expressed complex as
E
=
σ
ε
=
σ
0
·e
i(ωt+δ )
ε
0
·e
i(ωt)
=
σ
0
ε
0
·(cos δ + i sin δ ) (Eq. 6.)
whereof E
0
= E
represents the storage and E
00
= E
the loss modulus. The ratio
denes the loss tangent δ
tan δ =
E
00
E
0
=
sin δ
cos δ
= tan δ (Eq. 6.)
Energy dissipated per cycle can be determined following the denition of the elastic
modulus
W =
I
σdε =
I
σ
˙
εdt (Eq. 6.)
The complex stress-strain relationship reads
σ(t) = E
0
·ε
0
·sin (ωt) + E
00
·ε
0
·cos (ωt) (Eq. 6.)
Inserting this in (Eq. 6.) with integration limits of one period yields
W
mc
=
Z
2π
ω
0
E
0
ε
0
sin (ωt) ·(γ
0
·cos (ωt))
dt
+
Z
2π
ω
0
E
00
ε
0
cos (ωt) ·(γ
0
·cos (ωt))
dt
= 0 + π ·E
00
·ε
2
0
(Eq. 6.)
The averaged dissipated power per cycle based on inter-molecular
˙
Q
mc
friction follows
the loss modulus term
˙
Q
mc
=
dW
dt
=
W
mc
2π
ω
=
ω ·ε
2
0
·E
00
2
(Eq. 6.)
75
6 Heating Models, Mechanisms and Parameters
with the oscillation frequency ω, strain amplitude ε
0
and loss modulus E
00
.
Besides the linear inuence of generator frequency f (ω), two points are from greater
importance: rstly, the quadratic dependency of dissipated power on the strain ampli-
tude representing the amplitude a
S,0
, and secondly, the linear dependence only on the
polymer loss modulus E
00
.
Thus, the amplitude is a sensitive instrument since a doubled amplitude creates four
times higher intermolecular heat dissipation. In addition, polymers with higher loss
moduli show better heating behaviours. Recalling the previous chapter, PPS matrices
showed much better weldability than PEEK in the pre-testing campaign. Following (Eq. 6.),
PPS should exhibit a higher loss modulus. Indeed, the loss tangent of PPS was found to
be up to one order of magnitude lower than the one of PEEK at room temperature (Ho
and Jow , ) hence E
00
PPS
E
00
PEEK
. Lower T
g
and T
m
are further indications for
this circumstance.
Benatar and Gutowski (8) proved a
strong temperature dependency of PEEK’s
loss modulus especially around T
g
(Fig-
ure 6.). Thus, the aim is to quickly es-
tablish zones with polymer temperatures
around T
g
by frictional heat to enhance
viscoelastic heating there inducing faster
melt front progression.
50 150 250 350
10
6
10
8
10
10
E”
E’
T
g
Temperature [°C]
Moduli [Pa]
Figure 6. Storage (E’) and Loss (E”) Mod-
uli for PEEK at  kHz (Data: Be-
natar and Gutowski 8)
6. Combined Heating Mechanisms
Ziegltrum (, -) investigated the inuence of static contact force for heating and
melting of thermoplastics with two main conclusions: ) below a certain applied con-
tact force there is no effective coupling between sonotrode and part increasing heating
time drastically, and ) generally leads an increase of contact force to an remarkable de-
crease in process time. This outcome is conrmed by obser vations of Villegas ().
76
6 Heating Models, Mechanisms and Parameters
For further increase of contact force,
Potente (, ) determined two re-
gions of heating according to Figure 6..
Below a distinct break point (blue) marks
the region of combined interfacial and
intermolecular frictional heating whereas
above (red), only intermolecular heating
was observed. Based on that, he devel-
oped a constitutive model for the simpli-
ed adiabatic heating process under ultra-
sonic oscillations as
(πη + 2mµ)E
ˆ
ε
2
f = ρc
dT
dt
(Eq. 6.)
with
m = 0.25
"
1
p
p
k
0.7
#
(Eq. 6.6)
0 500 1,000 1,500
0.1
1
a
m
=. µm
a
m
=. µm
a
m
=8. µm
a
m
=.8 µm
a
m
=6.8 µm
Static Contact Pressure [MPa]
Melting Time [s]
Figure 6. Melting Time of PMMA over Con-
tact Pressure for Different Ampli-
tudes (Data: Potente , )
using damping constant η, friction coefcient µ, Young’s Modulus E, amplitude strain
ˆ
ε,
oscillation frequency f , polymer density ρ and specic thermal capacity c as well as
static pressure p and pressure at break point p
k
.
In case of solely intermolecular friction, the facial friction term 2mµ disappears. Above
T
g
, the shear heating term τ
˙
γ replaces it.
Thereout, one can conclude that interfacial friction gets redeemed since relative move-
ment necessary for frictional heating is more limited with increased contact pressure
and eventually no relative motion between adherends is possible. By this, contact pres-
sure shall be chosen high enough to ensure good transmission of energy throughout
vibration application, but low enough to avoid limited relative motion of the parts. This
includes the xture of parts.
77
6 Heating Models, Mechanisms and Parameters
6.6 Multi-Body Dynamics and Inter facial Friction
The oscillating system of ultrasonic welding set-up can be represented by a serial spring--
damper system (Benatar and Gutowski 8). Energy loss per cycle in a damper under
harmonic oscillation follows (Stutts )
W
d
=
I
F
d
dx =
I
D ˙xdx =
I
D ˙x
2
dt = Dω
2
x
2
0
Z
2π
ω
0
cos(ωt)dt = πDωx
2
0
(Eq. 6.)
with the respective damping coefcient D. For an increasing number of masses with as-
sociated springs and dampers, energy losses must be summed up. Depending on the
damping system, this dissipated energy is not implicitly involved in the heating mecha-
nism.
In addition, (Eq. 6.) shows an indirect proportionality of dissipated friction heat and
applied surface. Either by inserting an additional loose layers and/or by deploying at
EDs, the surface is increased drastically by factor two or more, causing less generated
friction heat important for initial heating due to less energy concentration and larger
heat conduction effects. This assumption is conrmed by the study of Villegas et al.
(). They investigated ultrasonic weldability of different ED shapes and conducted
their study with triangular EDs directly moulded on the substrate, triangular EDs moulded
on a loose resin stripe and a at ED lm (Figure 6.). The characteristic power-dis-
placement curves revealed fastest initial heating for EDs on the substrate, followed by
moulded resin stripes and far behind the at ED lm. However, the nal cycle time dif-
fers since an interaction of all heating mechanisms takes place after initial melting and
during adhesion development.
(a) Triangular ED moulded
on Substrate
(b) Triangular ED moulded
on Loose Resin Stripe
(c) Flat ED Film
Figure 6. Investigated ED Forms
BRANSON follows previous elaborations for ultrasonic welding of three separate parts:
in a two-step process, input energy is concentrated on one interface after the other,
rstly joining the upper and middle part, and secondly, the joined with lower part.
Lastly, each interface layer does not only provide an additional surface but afliated
transmission losses further decreasing the process efciency.
78
6 Heating Models, Mechanisms and Parameters
6. Composite Heat Flow Behaviour
Since composites exhibit no isotropic material behaviour, basic law of heat conduction
F’ L (Baehr and Stephan 6, )
˙q = λ ·T (Eq. 6.8)
must be rewritten as
˙q
i
= λ
i j
·
T
x
i
with i, j = (1, 2, 3) (Eq. 6.)
with the heat conductivity λ
i j
as nd order tensor. Assuming the unit cell method (Thomas
et al. 8), the tensor reads
λ
i j
=
λ
11
0 0
0 λ
22
0
0 0 λ
33
. (Eq. 6.)
For the given /° fabric, λ
11
and λ
22
represent bre directions (λ
k
), λ
33
the thickness
direction (λ
). The latter can be determined by applying rules of mixture
λ
= ϕ ·λ
f
+ (1 ϕ) ·λ
m
(Eq. 6.)
with the bre volume content ϕ and thermal conductivities of bres and matrix λ
f
=
10.46
W
/m ·K and λ
m
= 0.29
W
/m ·K, respectively. Values are taken from the data sheets of
deployed materials (see Appendix A). Fibre volume content in aerospace applications
typically ranges between . .6. For the given laminate, ϕ is denoted as ..
For a rough approximation, thickness direction can be modelled as a at composite
wall consisting of layers of bres and matrix, alternately. Occurring stationary heat ow
˙
Q can be generally described expanding (Eq. 6.8) by an arbitrarily chosen area element
dA to
d
˙
Q = λ ·T dA. (Eq. 6.)
Considering a one-dimensional heat ow in thickness direction z reads
d
˙
Q
z
= λ
z
·
T
z
dxdy. (Eq. 6.)
79
6 Heating Models, Mechanisms and Parameters
Integration, rearranging and anew integration yields
˙q
z
=
λ
z
t
·T = k ·T . (Eq. 6.)
Heat conductivity in thickness direc-
tion can be computed according to a se-
rial arrangement of layers (Figure 6.). For
simplication, heat transfer between lay-
ers is assumed as ideal hence α
i
= α
o
. With the bre volume content, relative
thicknesses are used nally getting the
thermal resistance
bre
matrix
bre
bre
matrix
bre
λ
f
λ
m
λ
f
λ
f
λ
m
λ
f
t
i
Figure 6. Composite Wall Model
k
z
=
1
n
i=1
t
i
λ
i
=
1
ϕ
λ
f
+
1ϕ
λ
m
=
1
0.5
10.46
+
10.5
0.29
W
m
2
·K
= 0.564
W
m
2
·K
, (Eq. 6.)
and the ratio of heat conductivity in and transverse bre direction
λ
k
λ
=
λ
f
λ
z
=
λ
f
k
z
·1
=
10.46
0.564
= 18.53 (Eq. 6.6)
In reality, ratio of melt front propagation in and transverse bre direction should exhibit
a lower value since coupling of bre and matrix hence heat transfer coefcient α
i
and
α
o
will provide a non-neglectable inuence. In addition, no stationary process will be
observed and heat transport towards laminate edges away from the heat source, i.e.
melt will be present.
Already at this stage, the anisotropic behaviour of composite materials marks several
peculiarities for later continuous process. Since a UD tape laying process is envisaged,
main heat ux will go away from the joint interface and will run ahead of the sonotrode
movement. Yet, heating in direction of motion must not be neglected and leads to a
pre-heated condition once the sonotrode will arrive at a later position on the joint line.
Due to this pre-heating, a reduced input energy is necessary for completion of the
welding process. Appropriate countermeasures are described by Senders (6, 6) as
halved amplitude and contact pressure. The latter contributes as well to a lower lateral
force during movement.
A FEM simulation of the given pre-heating scenario is worth considering.
80
6 Heating Models, Mechanisms and Parameters
6.8 Theory Conclusion
Respective dependencies of input, frictional and intermolecular power dissipation as
well as enthalpy of fusion from the previous considerations are summarised in the fol-
lowing Table 6., giving an indication of the form of proportionality by the schematics.
Table 6. Parameter Inuence
Frequency Amplitude
Contact
Force
Surface
Roughness
Contact
Area
Loss
Modulus
f a
S,0
F
C
R
a
A E
00
H
m
x
˙
Q
in
x x x
˙
Q
f ric
x x x x
/x
˙
Q
mc
x
x
x
˙
Q
f ric
/
˙
Q
mc
8:
6 Heating Models, Mechanisms and Parameters
6. Scientic Approach
During the parametric study (Chapter 8), the discussed and below elucidated parame-
ters shall be adjusted in a way to maximise joint quality, weld strength and joint exten-
sion/melt front propagation.
The operating frequency is constant at  kHz; frequency uctuations are recorded
and determined distinctively under % of the operating frequency.
The amplitude varies for each weld since the microprocessor determines the most
efcient power-energy-ratio and adjusts the peak power (linear proportionality to ampli-
tude) automatically.
The contact force shall be a direct input parameter changed in the generator’s user
interface.
The surface roughness is highly depending on the manufacturing process. For in-
uence investigations, hot press and vacuum consolidated laminates are observed ex-
hibiting different nish roughness due to the varying processes.
The projected contact area remains constant since only at energy directors are de-
ployed. However, the ED dimensions and thicknesses shall be varied.
The loss modulus is solely depending on the material used. In this case, all materials
shall remain the same throughout the experiments.
Despite elaborated inuences of amplitude and contact force, in accordance with typ-
ical parameters in comparative studies, the parameter set shall be complemented with
the consideration of energy input (Senders 6; Villegas and Palardy ) and the ED
lm thickness (Palardy and Villegas ; Senders 6).
However, input energy represents a certain amount of energy for the given set-up. In
view of later continuous application, input time calculated out of the ratio of input
energy and power can be more easily transferred into a longitudinal motion since
the feed velocity will be the crucial process parameter in that case considering the
effective duration of ultrasonic oscillation during movement.
82
Experimental Set-Up
. Material
For more detailed data extracted from respective data sheets of used materials, please
refer to Appendix A.
.. Laminate
Unlike later tape laying application with UD ply stack up, °/° fabrics were acquired
for this parametric study providing better comparability with respective studies.
The acquired product are Carbon Fibre (CF)/Polyetheretherketone (PEEK) plates by
H C. Wrought material is a °/° fabric semi-prepreg (ve harness
satin) consisting of T HT-Carbon bres (K) and Victrex
®
G matrix. Seven plies
of [/] fabrics are consolidated in a hot press process to plates of 6 x  mm
with a nal bre volume content around  %.
During manufacturing, a release lm was used in the hot press and removed after-
wards. Fur thermore, all specimen are cleaned with acetone immediately before welding
to remove release agent or grease debris from handling.
A second batch of plates was produced with same materials mentioned above, but
with a vacuum consolidation process at DLR Augsburg.
.. Energy Directors
Energy directors are either an APTIV
®
 black neat resin lm (soot-blackened) based
on Victrex
®
PEEK with a thickness of  µm or a LITE
®
TK  µm PEEK lms.
In case of weldings with  µm ED thickness, two LITE
®
TK lms are placed loose
above each other and positioned via a spot weld xation (Figure .) created with a hand-
held BRANSON  kHz LPe :.:T unit and rectangular sonotrode of ." x ."
(. x .8 mm).
.. Test Specimen Design
To ensure comparability of test results, ASTM D  is utilised as standard for test
specimen design. Intensive studies on CF/PPS and CF/PEI ultrasonic welding were con-
7 Experiment al Set-Up
ducted by Senders (6), Villegas et al. (), Villegas and Palardy () or Villegas
() using this standard type providing a good database for later evaluation and com-
parison.
Test specimen dimensions can be taken from Figure .. It should be noted, that
thickness is increased from .6 mm to . mm due to manufacturing reasons of test
plates. Overlap is only a recommended length in the standard and harmonised with
existing test set-ups. Further information extracted from ASTM D  can be found in
Appendix A.
.6
L=.
.6
.
at ED (L+ 6 mm)
Gripping Area
Gripping Area
.
. 6. . 6. .
.
Figure . Test Specimen Dimensions acc. to ASTM D 
Specimens were produced out of the delivered test panels via water jet cutting op-
eration offering a high quality and least impact on the material, e.g. by overheating or
mechanical damage.
The cutting method provided test specimen of hot press processed parts with a °/°
bre orientation and vacuum consolidated specimen with a ±° bre orientation, re-
spectively.
. Manufacturing Equipment
.. Ultrasonic Welding Machine
All tests are conducted with BRANSON System Xd with a peak power of  W.
The generator works with an operating frequency of  kHz and provides a maximum
number of 8 cycles per minute. The integrated feed drive aec . with a " pneu-
matic cylinder offers a maximum weld force of .6 kN and a dynamic trigger range from
84
7 Experiment al Set-Up
 N upwards (Figure .). The microprocessor controlled machine opens the eld for
time, energy and displacement controlled applications (BRANSON ).
.. Sonotrode
The deployed sonotrode is an OF-886 rectangular steel sonotrode with a planar con-
tact surface of 6 x 6 mm. It is combined with a gold booster for a :. transmission
and a nal peak-to-peak amplitude of 8 µm, measured with a dial gauge. Planarity of
the sonotrode on the specimen was veried with a sheet of white and carbon paper.
When pushing the sonotrode manually against inserted sheets on the specimen, the
carbon imprint indicates the necessary adjustments of the machine table in order to
achieve a plain contact interface.
.. Fixture
Two xture solutions are deployed: a machine modied item
®
prole 8x mm in
heavy design with 8 mm groove width and  mm groove distance, respectively. The
T-slots provide a variable clamp distance. Clamping jaws were designed individually for
the application and manufactured out of steel. For variability and concerning the hole
pattern of the testing machine environment, a second prole was introduced offering a
6° mountability of the xture (Figure .a) at location, no use of second prole was
necessary.
(a) Machined item
®
Prole (b) Steel Clamping Table
Figure . Fixture/Anvil Variants
The second xture reuses the clamping jaws in combination with a steel clamping
table as anvil (Figure .b), provided by BRANSON.
Additional steel plates and peek lm patches are used to level the set-up for a planar
contact surface between sonotrode and upper adherend as well as between upper and
85
7 Experiment al Set-Up
lower adherend (Figure .).
Screws of the clamping jaws are tightened with a torque wrench to maintain, on the
one hand, a comparative xation throughout experiments and, on the other hand, to
avoid huge asymmetries in longitudinal or lateral direction hence introduced pre-stress.
Figure . Test Stand
Figure . Ultrasonic Prexation of Two
Loose PEEK lms on Lami-
nate with Handheld Unit
Figure . Steel Table Anvil
86
7 Experiment al Set-Up
. Analysis Methods
The used methods for the evaluation and discussion of the test results are described in
the following sections.
.. Manual Bending/Breaking
At the beginning of the experiments, test specimen are bended and broken manually
(Figure .6) to investigate the degree of melt fronts at the fracture surface for adjusting
the process parameters towards more effective parameter sets and congurations.
Figure .6 Manual -Point-Bending Test
.. Lap Shear Tension Test
The specimen design is chosen in accordance with ASTM D  to provide standard-
ised lap shear tests with representative strength results. For this purpose, the ZWICK  ten-
sion testing machine is used with a  kN load cell and a traverse path sensor, located
at the DLR Institute of Structures and Design, Stuttgart.
.. Fracture Surface Analysis
For manually as well as automatically tested and destroyed specimen, fracture surfaces
are carefully investigated offering valuable information about melt front propagation or
unwelded areas. Visual inspection with naked eye shall be performed as well as by
digital microscope using the K VHX-.
87
8 Parametric Study, Results and Discussion
8. Anvil Stiffness
During rst welding experiments with the aluminium item
®
prole, no or only very little
joining was achieved. After changing to a steel machine table as anvil, joints are estab-
lished much more easily. The hypothesis, the aluminium prole shape and material do
not provide sufcient stiffness for ultrasonic oscillation introduction, arose.
This theory can be veried with an ANSYS harmonic response analysis of deployed
xtures (Figure 8.). The operating frequency is varied between  and  Hz
(corresponding to the actual measured frequency during testing), the introduction area
is a box representing actual specimen thickness and joint position. Fixed supports are
set at the bottom of item
®
brackets and machine table screw slotted holes, respectively.
The bottom surfaces are restricted with a remote displacement in motion in vertical
direction.
The simulation results show an over six times larger maximum total deformation for
the aluminium item
®
prole (6.µm) compared to the steel table (.µm). In addition,
maximum deformation occurred near the introduction area hence in the direct range
of inuence for test specimen (Figure 8.a). Contrary to that, lower deformation of the
machine table occurred far away from sonotrode position and at a much smaller extent.
(a) Aluminium item
®
Prole
(b) Steel Machine Table
Figure 8. ANSYS FEM Harmonic Response Simulation for Total Deformation of Deployed
Anvil Variations at a Frequency Range of  to  Hz;
Point of Oscillation Introduction is chosen at the Actual Specimen Position with the
used Sonotrode Area (Block)
Consequently, input energy by the ultrasonic unit is transferred at a higher degree into
8 Parametric Study, Results and Discussion
deformation work for the aluminium prole. This energy is missing at the joint interface
for heating and melting of adherends hence inferior ultrasonic welding performance.
Another disadvantageous material proper ty of aluminium is the by two orders of a mag-
nitude lower damping factor (Beards , ) and an almost pure elastic behaviour of
steel (Ehrenstein et al. , 8).
Concluding, the anvil must exhibit sufcient stiffness determined by material and
shape to enable ultrasonic welding. Villegas and Bersee () followed the approach
of a combined aluminium xture with steel anvil from solid blocks (Figure 8.).
Figure 8. Combined Aluminium-Steel Solid Fixture Design (Villegas and Bersee )
8. Specimen Arching
After clamping and before welding, depending on the clamping screw torque, the spec-
imen exhibited an arched position according to the schematic in Figure 8.a. The lower
adherend shows this as well but at a smaller extent.
The higher the torque chosen, the higher the induced compression stresses hence
tilting. Since a rm clamping is required, reduction of the torque can only be achieved
until a lower threshold.
As a containment action, a steel plate is inserted to compensate the arching at the
outer specimen side and reduce the deection at the joining interface (Figure 8.b).
(a) Schematic Tilting
(b) Steel Insert Plate
Figure 8. Clamping-Tilting Issue
89
8 Parametric Study, Results and Discussion
Despite aforementioned countermeasure, still some arching occurs which can inu-
ence the welding process dramatically. The ultrasonic welding machine works with a
trigger force which compacts the stack before welding cycle. Once the trigger force is
reached, the oscillation starts. Are the induced compression stresses little too high and
trigger force in the sonotrode is reached before the deection is compensated means
adherends contact each other, the microprocessor could misinterpret recorded data and
aborts the process. Is the pre-stress way too high, arching cannot be compensated and
sonotrode oscillation sets off vibrating upper adherend only.
Therefore, it must be ensured that arching does not occur in an excessive manner
via lower torques or closer clamping position. In addition, trigger force must be set
high enough to star t oscillation when adherends are in contact preferably planar
otherwise leading to edge effects.
8. Edge Effects
Edge effects shall describe melt initiation at edges of the specimen. Edge effects at
sidewards positions are solely a phenomenon on specimen size. For a continuous ap-
plication, sideward edges disappear” along the path of sonotrode motion due to large
dimensions and must only be considered at the run in and run outs.
Edge effects shall be clustered into three groups of causes.
8.. Unbalanced Clamping Lateral
The preceding section described the case of unbalanced clamping around the lateral
axis resulting in specimen arching. In case pre-stress is under a certain threshold, the
welding process is executed, though.
However, no planar contact situation is
present but an initial line contact at the in-
ner edge of the lower adherend indicating
a higher degree of curvature of the upper
specimen at the beginning of the process.
Frictional contact hence melt initiation at
this side can be observed (Figure 8.).
Figure 8. Initial Line Contact
Over time with creation of melt, displacement of the sonotrode downwards sets in
pushing the wedge-shaped melt front proceeds towards the other edge (Figure 8.).
The input energy limits the extent of molten resin.
90
8 Parametric Study, Results and Discussion
(a) D scan (b) D detail view
Figure 8. Melt Flow from Inner Edge due to Initial Line Contact
8.. Unbalanced Clamping Longitudinal
Not only tilting around the lateral, but also around longitudinal axis occurs. If screws are
not alternately tightened or with slightly different torques, higher contact pressures on
different sides are achieved. Those lead to higher frictional energy dissipation than at
areas with lower impact. The melt initiation will most likely start from this edge instead
of evenly distributed over the welding area (Figure 8.6).
(a) higher clamping force on the right (b) higher clamping force on the left
Figure 8.6 Sideward Edge Effect due to Longitudinal Unbalanced Clamping
A case of combined lateral and lon-
gitudinal unbalanced clamping can oc-
cur, too. Consequently, not only one
edge shows an initial melting spot
but an area expanding over the corner
along two perpendicular edges on the
side of higher areal loads (Figure 8.).
Figure 8. Combined Lateral/Longitudinal
Unbalanced Clamping Edge Effect
9:
8 Parametric Study, Results and Discussion
8.. Edge Concentration Conditions
In general, edges inherit high potential for irregularities induced during manufacturing.
Either the cutting of the specimen or insufcient deburring leads to little peaks acting as
primary energy director at the initial stage. Frictional heat generation known as process
starter focus on those points rather than an areal motion and melting.
The specimen in Figure 8.8 shows preferential melting on either sides speaking against
unbalanced clamping towards one side. The D-scan reveals expected edge irregulari-
ties induced before welding stage.
(a) top view (b) D scan
Figure 8.8 Preferential Heating due to Edge Unregularities
8. Patch Approach
Since aforementioned edge effects had been discovered at an early stage, original ex-
perimental set-up was changed from areal deployment of at ED (PEEK lm) stepwise to
strips omitting front and back edges and nally to patches placed centred on the lower
adherend. Volkov et al. (a; b) already proved the inuence of surface micro
irregularities on later welding quality.
The achieved effect is simple: by in-
serting material patches only in the cen-
tral area, the outer edges are lifted by
a small but sufcient amount to avoid
friction concentrations at irregularities on
the outer edges hence edge effects (Fig-
ure 8.).
Figure 8. Lifting Effect of ED Patches for
Avoiding Edge Effects
92
8 Parametric Study, Results and Discussion
Further was discovered a lack of ma-
trix material at the interface with  and
 µm PEEK lms by reference to dry -
bres on fracture faces and no matrix de-
bris (Figure 8.). This was compensated
with insertion of two  µm loose lms
on top of each other with the need to ultra-
sonically pre-x the at EDs (acc. to Sec-
tion ..) as experiments showed already
shifting of strips during oscillation phase.
Figure 8. Dry Fibres Indicate Lack of Ma-
trix after Deployment of  µm
PEEK lm
The thickness of deployed at ED lms is strongly dependent on the thickness of
unreinforced surface layer. For thinner matrix top layers, more additional resin must be
inserted in form of thicker ED lms and vice versa.
Those described adjustments let arose two phenomena elucidated hereafter.
8.. Guided Melt Initiation
Several fracture surfaces indicate a preferential heating and initial melting near the ultra-
sonic pre-xation joint. Figure 8.a shows the initial stage of melting at and near by the
pre-xation. Figure 8.b proves that even where the ED’s lower right corner exhibited
best circumstances for melting, the pre-joint started melting at a completely different
position, too. Figure 8.c represents a far molten state with a direction of propagation
from the centre (point of pre-xation) towards the outer edges.
(a) (b) (c)
Figure 8. Fracture Surfaces with Indication of Guided Melt Initiation
Although it is not possible to associate a certain amount of contribution to the pre-x-
ation joint, the experiments showed evidently that melting is more easily initiated when
a pre-joint is present.
93
8 Parametric Study, Results and Discussion
Reasons for this behaviour are various: the pre-created material bonding acts as start-
ing point, the thinned ED lm at the pre-joint requires less fusion enthalpy or the irregu-
larity of the joint shape in the at ED caused friction concentration.
Whether one or several of the listed factors inuence the eased melting should be
further investigated.
8.. Interfacial Friction
The deployment of two loose ED lms
arises the issue of interfacial friction dur-
ing oscillation phase. Although xed at
the pre-joint point, the outer areas are still
able to execute relative motion between
the two layers which creates an additional
heat source. Indeed can such interlayer
melting be observed (Figure 8.). The
line pattern of brighter and darker areas
indicates unmolten and molten material
shining through the lm surface. As the
top sur face is still in an unmolten state,
melting must be occurred between the lay-
ers. The small extent of melt fronts con-
rms the theory of reduced efciency for
multiple layered loose resin lms (Sec-
tion 6.6).
Figure 8. Interlayer Melting Effect
8. Failure Modes
Failure modes for bonded composite joints are generally dened as ) adhesive failure
(interface adhesive/adherend), ) cohesive failure (inside adhesive layer) and ) sub-
strate failure (Vassilopoulos , ). Strong et al. () expanded them to four fail-
ure modes for ultrasonic welding: ) weld inter facial failure in resin-rich areas, ) com-
bined interlaminar and interfacial failure, ) interlaminar failure above and below energy
susceptor layer and ) coupon failure due to bre damage (no weld failure).On that base,
fracture surfaces are analysed.
Since all specimen were prepared with acetone to clean fusion surfaces and remove
greasy debris, typical failure mode of bad surface preparation adhesive failure was
94
8 Parametric Study, Results and Discussion
not expected and indeed not obser ved.
Generally, welding quality exhibits good to ver y good properties. Characteristic cohe-
sive failure for given application with tore interlayer material (Figure 8.a) is found as
well as even some bre pull-outs on specimen indicating Strong’s failure modes and ,
i.e. coupon failure hence stronger bond than substrate properties (Figure 8.b). Some
specimen show typical cohesive failure combined with bre pull-outs created due to -
bre re-orientation during intensive melting in thickness direction (Figure 8.c). Then,
upper plies get detached from substrate and “oat” towards interface layers such an
effect is undesired since intended bre orientation of top layers gets lost.
(a) Cohesive Failure with Tore
Matrix
(b) Substrate Failure
(Pulled-Out Fibres)
(c) Cohesive Failure and Fibre
Pull-Outs at Weld Bead
Figure 8. Observed Failure Modes on Fracture Surfaces
8.6 Heat Flow Behaviour
8.6. In-Plane Direction
The experiments show a clear dependency of top layer bre direction on the melt front
propagation. Section 6. already introduced the theoretical model based on that heat
ow along bres is at least one order of magnitude higher than transverse. Subse-
quently, bre direction plays a major role in heat and thus melt propagation.
Figure 8.a and 8.b show evidently melt directions of laminates with ±° and
°/° top layers, respectively. Therefore, top layer bre orientation can be a useful
adjusting screw in combination with sonotrode geometry and motion as well as ED po-
sitioning to guide the meld ow. Since a UD tape laying process is envisaged, this tool
might be even more powerful.
To note is the distinct higher edge heating compared to the surface temperature due
to preferential bre heat ow. Thus, squeezed out matrix contacting laminate edges is
likely prone to overheating hence matrix decomposition and must be avoided.
95
8 Parametric Study, Results and Discussion
(a) ±° Top Layer with Corresponding Melt
Direction
(b) /° Top Layer with Corresponding Melt
Direction
Figure 8. Observed Failure Modes on Fracture Surfaces
8.6. Thickness Direction
The considerations in Section 6. do not only point out the role of bres as major heat
transfer medium, but does also give an estimation of heat propagation in and transverse
bre direction.
Experiments prove theoretical predictions and exhibit a typical triangular melt front
shape (Figure 8.). Further, the assumption of reduced ratio of heat conductivities
compared to theoretical values is valid, too, via graphical determination of melt front
propagation yielding
λ
k
λ
d
k
d
=
11.5
1.25
= 9.2, (Eq. 8.)
compared to the theoretical rough estimation (18.53) from (Eq. 6.6), the actual ratio is
quite accurate half of the theoretical value.
Further evidently is no impact or even melting on possible outer aerodynamic sur-
faces opposite joining interfaces although vibration times were comparably long in the
conducted study.
d
= 1.25 ul
d
k
= 11.5 ul
ul: unit length
Figure 8. Microscopical Analysis of Melt Front Propagation In/Transverse Fibre Direction
96
8 Parametric Study, Results and Discussion
8. Parametric Study
Finally, after adjusting boundary conditions towards a better and repeatable welding
quality, a test series of in total ten specimen had been conducted. Results and interpre-
tations are elucidated in the following sections.
8.. Obtained Experimental Data
The obtained experimental data (Table 8.) consist of recorded US machine data (Am-
plitude, Energy Input, Weld Collapse/Force/Time, Frequency), tension testing machine
data (F
max
), microscopical determined fracture surface area (A) and computed values
of mean power, energy density and lap shear strength (LSS).
One should note that determination of fracture surface area is crucial for LSS calcula-
tion. The denition of bonded area measured is ambiguous and shall be characterised
as molten unreinforced top layer on the opposite adherend of the one with pre-xed EDs.
This guarantees measured areas being involved in the joint whereas molten ED/matrix
only on the “pre-xed” adherend does not explicitly have to form a joint but melt on one
of the laminate.
Although technical aids of the digital microscope are used to ensure a systematic
measurement, still considerable deviations among several measurements of the same
specimen are observed. In addition, the set denition of bonded area still leaves room
for interpretation, e.g. haze caused by matrix debris.
Statistical evaluation has been carried out according to the procedures presented in
Appendix B.
8.. Force-Elongation-Curves
The force-elongation curves cannot provide an overview of joint strength since reassess-
ment to the respective area must be done. Nevertheless, they display the failure be-
haviour over time. A summary of all valid specimen curves is presented in Figure 8.6.
Most specimen exhibit characteristic curves with a distinct maximum (e.g. #, #,
#8). By contrast # shows a stretched and # clinched shape. Regarding respective
fracture surfaces, it is evident that # shows a rather random compartmentalised dis-
tribution of molten matrix (Figure 8.a) contrary to the joint area of # with its main
extent in direction of load (Figure 8.b).
97
8 Parametric Study, Results and Discussion
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
0
500
1000
1500
2000
2500
3000







6
8
Traverse Path [mm]
Force [N]
Figure 8.6 Force over Traverse Path (with Specimen ID)
(a) # (b) #
Figure 8. Fracture Surfaces with Marked Joint Area
Stress analysis investigations detect maximum peel stresses at outer edges of single
lap shear specimen with a parabolic distribution over the overlap length (Figure 8.8).
Thus, one can conclude the shorter the actual overlap/joint length in load direction is, the
higher is the peel stress gradient in particular for compartmentalised formations. A
decrease in overall shear strength leads in turn to an earlier failure and plasticity effects
hence higher elongation does not appear.
This theory is further conrmed since experimental data show a similar trend (Fig-
ure 8.). Smaller weld areas lead to lower LSS whereas innitely large joint areas ap-
proximate asymptotically substrate strength hence ideal bond. Two outliers mark the
highest and lowest measured values accounting for systematic errors in area determi-
nation.
98
8 Parametric Study, Results and Discussion
Figure 8.8 Peel Stress Distribution
(FAA , 8)
0 20 40 60 80 100
0
5
10
15
20
25
30

8
Joint Area [mm
]
LSS [MPa]
Figure 8. LSS over Welded Area
Microscopical pictures of all specimens fracture surfaces with marked joint areas are
provided in Appendix C.
8.. Vibration Time-Force-Correlation
Plotting vibration time over applied
weld force exhibit whilst with consid-
erable deviations a linear growth (Fig-
ure 8.) and thus resembles the theo-
retical model by Potente (, ) pre-
sented in Section 6.. Albeit, from the
obtained data, one cannot ascertain be-
yond doubt in which of the two regions
(combined/solely intermolecular heating)
the experimental data lies.
One should note, that different weld
forces occurred for pneumatic cylinder
pressure held constant. The need for
a more precise electro-mechanical posi-
tioning unit becomes evident.
1775 1780 1785 1790 1795
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Weld Force [N]
Vibration Time [s]
Figure 8. Vibration Time over Weld Force
99
8 Parametric Study, Results and Discussion
8.. Vibration Time-Amplitude-Correlation
The graph of vibration time over ap-
plied amplitude provides the expected hy-
perbolic shape (Figure 8.) validating the
derivation of (Eq. 6.) from Section 6.
with its quadratic dependency on ampli-
tude. Remarkable are two asymptotes:
vertically marking the minimum amplitude
required for welding. Horizontally, the
minimum vibration time with a decreas-
ing gradient for larger amplitudes. This
state of equilibrium results due to multi-
plication of increasing amplitude but drop-
ping loss modulus above T
g
(refer to Fig-
ure 6.) avoiding a further reduction in vi-
bration time.
5
10 15
1.0
1.5
2.0
2.5
Peak Amplitude [µm]
Vibration Time [s]
Figure 8. Vibration Time over Weld Force
Note: variation in amplitude due to US welder am-
plitude automatic for optimum power.
8.. Weld Area-Collapse-Correlation
The linear increase of weld area with
greater weld collapse (Figure 8.) can
be explained as more weld collapse rep-
resents more molten material hence in-
creased melt front propagation, i.e. larger
welded area. Effects of squeeze-outs af-
ter melting of entire overlap area would
form a knee in the graph since collapse
rises but welded area remains constant.
Specimen revealed melt propagation only
within the overlap due to patch approach
hence no such indications.
0.20 0.25 0.30 0.35
20
40
60
80
100
Weld Collapse [mm]
Weld Area [mm
]
Figure 8. Vibration Time over Weld Force
:00
8 Parametric Study, Results and Discussion
8..6 LSS-Energy Density-Correlation
The energy density is computed as
quotient of input energy and welded area
and is thus comparative to enthalpy of fu-
sion. In Figure 8. an optimum process-
ing window (– J/mm
)can be deter-
mined. A lower threshold similar to the
fusion enthalpy marks a certain energy
density that must be reached to enable
melting. For higher energy densities, LSS
decreases due to higher thermal burden
for the matrix material and reduction of
its properties. Above an upper thresh-
old, matrix decomposition would set off.
For some specimen and conguration en-
hanced smoke emission was observed in-
dicating matrix decomposition.
0 10 20 30 40
0
5
10
15
20
25
30
Energy Density [J/mm
]
LSS [MPa]
Figure 8. LSS over Energy Density
8.. LSS-Mean Power-Correlation
Mean power is the ratio of input en-
ergy and vibration time. The curve in Fig-
ure 8. exhibits again a distinct region
for maximum LSS. Akin to the preceding
curve for energy density, a lower threshold
marks the minimum power required for
melting, an upper threshold setting in ma-
trix degradation with dropping LSS. The in-
-between window (– W) provides
best boundary conditions for good weld-
ing quality.
300 350 400 450 500 550 600
0
5
10
15
20
25
30
Mean Power [W]
LSS [MPa]
Figure 8. LSS over Mean Power
:0:
8 Parametric Study, Results and Discussion
8..8 LSS-Vibration Time-Correlation
Corresponding to mean power from
previous consideration and since P =
E
in
/t
for constant input energy, a similar pro-
cess window opens for LSS over vibration
time according to Figure 8. between .6
and . s. Villegas and Palardy() ob-
served the same curve for CF/PPS spec-
imen as well as Strong et al. () for
APC- materials. In addition, suitability to
control the process by power, input energy
or vibration time is proven.
1.2 1.4 1.6 1.8 2.0 2.2 2.4
0
5
10
15
20
25
30
Vibration Time [s]
LSS [MPa]
Figure 8. LSS over Vibration Time
8.. Comparative Lap Shear Strength / Weld Factor
For determination of a comparative lap shear strength, a base line is necessary from
which the weld factor can be computed. The latter is referred to as “... ratio of weld
strength to strength outside the welded zone, typically determined by tensile stress
tests. (PDL 8, ) Hereby, interlaminar shear strength according to DIN /68
can be used as well as the shear strength of the parent material in composites the
weaker matrix.
Tests were conducted following
DIN 68. Yet, only three out of nine
specimen exhibited a shear failure mode
(Figure 8.6) and computed results ex-
hibited disproportionate and deviating
results. Main disturbing factor was
localised as notches cut by hand and
missing guiding devices due to too little
specimen.
Figure 8.6 Torn Interlaminar Lap Shear
Specimen
As a consequence, weld factor is determined with the parent material properties.
Laminate data sheet provide a shear strength of  MPa. Recalling the weld factor
denition with given values reads
w
US
=
σ
w
σ
p
=
27.8
53
= 0.52. (Eq. 8.)
:02
8 Parametric Study, Results and Discussion
Table 8. gives an overview of achieved experimental results and a comparison with
quantities of former publications and material combinations. According to this selec-
tion, the achieved welding quality shows the best weld factor and is among the best
absolute weld strength values.
Table 8. Weld Strengths and Factors for Material Combinations
Villegas 
Villegas and
Bersee 
Villegas
et al. 
Senders 6
Silverman and
Griese 8
Taylor and
Jones 
Experiments
CF/PEI CF/PPS APC- (CF/PEEK laminate)
σ
p
[MPa]   8. 8.   
σ
w
[MPa] . .-6. . .-6.  .8
Weld Factor .6 .-. . .-. .6 . .
Nevertheless, most recent publications investigated CF/PEI and CF/PPS, respectively.
The database for ultrasonically welded CF/PEEK composites contains rather little and
rather old state of work. Hereby, this work is another step to get an insight in an ultra-
sonic welding process with CF/PEEK laminates.
8.8 Conclusion
This conclusion shall lay the foundation for subsequent considerations on a continu-
ous ultrasonic welding process and provide the necessary background for design and
dimensioning.
For successful implementation, anvil shape and stiffness turned out to be of great
importance. Even more severe appeared adherends planarity. Induced (lateral/longi-
tudinal) arching or edge irregularities concentrate ultrasonic vibrations on single spots
rather than the entire overlap area causing uncontrollable melt initiation and propaga-
tion. Countermeasures like steel inser t plates and ED patch approach were deployed
successfully and emphasise the importance of planarity once more. Adjustments in
sonotrode geometry are considerable, too. Main challenge is assigned to achieving an
areal weld hitting the process window between “no welding” and “local overheating/de-
composition for an equal melt propagation.
Preferential heating near ultrasonic pre-xations was observed as well as interfacial
melting between two loose ED lm. The latter is a consequence of a lack of matrix at the
:03
8 Parametric Study, Results and Discussion
joint inter face and doubling of inserted ED lms. In combination with the unreinforced
top layer of the laminates, this is an important adjusting screw for enough matrix at the
interface. Only then, failure modes appear to be cohesive or even substrate failures like
in the present case.
Further conrmed is the preferential heat ow in bre direction which will arise an
issue for the continuous approach over large lengths with its pre-heating phenomenon.
Fibre orientation in the top layers sets another adjusting screw together with sonotrode
geometry and motion as well as ED positioning.
Parametric study revealed good to very good welding qualities with comparable high
LSS, weld factors and satisfying parameter correlations. Yet, vibration times and weld
force turned out to be higher as usual or desired, respectively. Moreover, melt propa-
gation i.e. welded area showed great deviations implying the need for further improve-
ments towards a more stable process. Energy density, mean power and vibration time
turned out to be crucial providing only certain process windows for maximum LSS and
dropping anges.
Finally, conducted experiments provide promising results, proves the feasibility of the
envisaged process and justies further investigations towards a continuous ultrasonic
joining process in this direction.
:04
Table 8. Obtained Experimental Data
ID
Ampli-
tude
Energy
Input
Weld
Collapse
*
Weld
Force
Freq
Vibration
Time
Mean
Power
F
max
A
Energy
Density
LSS
µm J mm N Hz s W
N mm
J/mm
MPa

. 8.8 .   .8 .6
. .6 .8 .66

. 8.6 . 86  . .
6. .6 .6 8.8

. 8. . 86  .68 6.
.8 6.6 8.88 .8

. 8. .8   .6 .6
.8 . .6 .

. 8. .   .66 8.6
8. . 6. 8.

. 8. .   . .
8. .68 8.6 .8

.8 8. . 8 6 . 6.
.8 .6 . .8
6
8. 8. .   . .
66. . . 6.6

. 8. . 6 6 . .6
- - - -
8
8. 8. .  6 .68 .
.6 .86 . .
Mean
8. . 8  . .
.
dev.
±. ±.8 ±6 ± ±. ±68.
±.
*
setting path/displacement
specimen broke immediately after welding whilst handling
Continuous Welding Concept Development
. Scope
The continuous welding process to be developed shall be derived from the application
case of longitudinal joining for two aircraft shells (Figure .). Therefore, the scope
comprises
establishing a continuous ultrasonic
fusion bonding process for
CF/PEEK composite shells
in overlap conguration
(- mm overlap) with
total joint length  mm,
preferably at energy directors,
fully-integrated endeffector for a
robotic device
applied in semi-automatic/
automatic mode.
Figure . Aircraft Shell Joining with Ultra-
sonic “Black Box” Endeffector
Combining these goals with the obtained experimental results from the previous chap-
ter shall be done in the following sections.
. Anvil
This point should regard the importance of a sufcient anvil stiffness as crucial prereq-
uisite for ultrasonic welding. Today, thermoset prepreg tapes are laid-up in female steel
tools (Figure .); similar to that, thermoplastic UD tapes would be stacked. For join-
ing and with compatible toolings, two shells can be positioned while remaining in the
lay-up tool using the latter as anvil. Advantages are absence of further toolings or trans-
fer stations hence time and cost saving plus suitable anvil usage. Disadvantage are
possible higher burden and wear of tooling surfaces necessary for outer aerodynamic
aircraft surface. Nevertheless, steel toolings for ultrasonic welding are widely spread
and combine good wear resistance with required hardness (Irshad , -).
9 Continuous Welding Concept Development
. Robot
One of the main goals is an automated joining process noted in the scope for a robotic
endeffector operating in semi-automatic and full-automatic mode. According to that,
the robotic system in particular robot and endeffector must be dimensioned regard-
ing collected experimental data. Contrary to stationary table machines which inherit
natural stiffness, in the given case, the robot arm must provide required stiffness be-
tween sontorde and anvil supported by its endeffector and the tooling.
The KR - PA is an existing robot at DLR Augsburg and chosen to be utilised for
the continuous joining test stand. The most crucial limiting parameters are both, the op-
erating range and the load capacity. “Both values (payload and mass moment of inertia)
must be checked in all cases. Exceeding this capacity will reduce the service life of the
robot and overload the motors and the gears [...]” (KUKA 6, ) The robot specica-
tion states a rated payload of  kg and a permissible moment of inertia of  kgm
.
The corresponding payload diagram for different endeffector weights is shown in Fig-
ure .. Given distances are related to the distance of the load centre of gravity with
respect to the mounting ange on robot axis 6.
0 100 200 300 400 500 600 700 800 900 1000
0
50
100
150
8 kg
 kg
 kg
 kg
6 kg
8 kg
 kg
L
z
[mm]
L
xy
[mm]
Figure . Payload Diagram for KR - PA (Data: KUKA 6, )
In case the required payload is not sufcient, an other robot must be chosen with a
higher payload capacity.
The load required is decisively inuenced on the one hand by the endeffector compo-
nents and its weight and on the other hand by the applied process loads, i.e. weld and
consolidation forces. Both are subject of the next section.
:07
9 Continuous Welding Concept Development
. Endeffector
The endeffector is the central unit for execution of an ultrasonic fusion process. Sev-
eral approaches have already led to patents for ultrasonic welding devices shown in
Figure ..
(a) (Gachnang 8) (b) (Soccard )
Figure . Existing Patents on Endeffector Concepts
Derived from those approaches and from observations during experimental study, the
endeffector must full the capability of
clamp/x the specimen in planar position,
incorporate US welding unit and
provide a consolidation device.
The necessity of an appropriate clamping/xation without any arching was intensively
discussed in the previous chapter. Installation of the ultrasonic welding unit in the end-
effector a fundamental prerequisite.
Senders emphasised the importance of a consolidation device during his investiga-
tions on continuous ultrasonic welding of CF/PPS: “... broken bres are shown, but also
large matrix akes and voids are present. [...] Therefore it is concluded that a consoli-
dation device is necessary in order to produce the same quality of welding as the static
welding process. (6, 6)
:08
9 Continuous Welding Concept Development
Following the principle of a positioning
device in front and a consolidation device
after the ultrasonic unit, the endeffector
concept in Figure . is developed. Fix-
ation and consolidation rollers frame the
sonotrode mounted independently. Thus,
different process forces can be applied
and varied. Major disadvantage is adding
up the process forces of each unit which
shall not exceed the robot’s payload ca-
pacity.
v
Adherends
Cons. US Fix.
Robot
Figure . Endeffector Concept
.. Process Forces
Total process forces are approximated as sum of used experimental forces.
Fixation was established with a torque of  Nm on the xture screws in correspon-
dence with former investigations (Senders et al. 6). Thereof, after analysing the ap-
plication case (Appendix D) with an chosen overlap width of  mm, the HERTZIAN pres-
sure yields a reasonable magnitude of clamping force in the range of  to  N. For
a conservative approach, the latter shall be used for calculation.
According to observations in experimental study, ultrasonic welding set in at weld-
ing forces of not less than 8 N, later parametric study was conducted with 8 N.
Comparable studies with CF/PPS used  and  N for static and continuous weld-
ing, respectively (Senders 6), also in order to lower lateral forces when transferring
the static into continuous approach. Thus,  N shall be the baseline for subsequent
calculation.
Consolidation is usually done with maintaining the welding force for a certain period
of time after end of heating stage hence F
weld
= F
cons
.
Thus, minimum required force capability for solely process forces reads
F
proc
= F
f ix
+ F
weld
+ F
cons
= F
f ix
+ 2F
weld
= 200 N + 2 ·1000 N = 2200 N m
proc
220 kg
(Eq. .)
KR - PA payload capability of  kg has not been exceeded in this estimation.
Still, there is additional weight impact due to deployed components.
:09
9 Continuous Welding Concept Development
.. Equipment Weight Impact
Apart from process forces, actual equipment brings in additional weight with every com-
ponent mounted. Components for a minimum working example (MWE) are given in Ta-
ble .. First three positions are pneumatic and electro-mechanical actuators responsi-
ble for applying process forces controlled by three load cells. The ultrasonic unit com-
prises the converter, booster, sonotrode and clamping device. Custom-made rollers,
coupling device and a framework round off the listing. If provided, weights are taken
from parts data sheets or are estimated according to comparable components. To
cover extra weight by additional devices and/or changes in purchased parts, a weight
allowance of  % percent is added to the subtotal.
Table . Equipment Weight Impact
Component Supplier Name F
max
l
S
Qty m
i
m
total
N mm g g
Fixation Act. FESTO ADN---A-P-A 8   
US Actuator FESTO ELGA-BS-KF--
-H-P-ML
   
Consolid. Act. FESTO ADN---A-P-A 8   
Load Cell HBM UC  -  
US Converter BRANSON  kHz CR-S - -  
US Booster BRANSON gold - -  
US Sonotrode BRANSON " x " - - 6 6
US Clamp BRANSON - - -  
Rollers - - - -  
Coupling Dev. - - - -  
Framework item Prole 8 x light - - .
kg
m 8
Subtotal 6
Weight Allowance  % 
Total Weight 68
By this estimation, total equipment weight impact is quantied to about  kg. Adding
to process forces from previous estimation reads
m
total
=
F
proc
g
+ m
Equip
=
2200 N
9.8065
N
/kg
+ 70.644 kg
= 224.34 kg + 60.804 kg = 285.14 kg
(Eq. .)
Thereby, robot’s payload capacity is not exceed, if only just. Nevertheless, once specic
::0
9 Continuous Welding Concept Development
planning is completed, executed estimation must be adjusted/completed and payload
capacity checked again.
.. Mass Moment of Inertia
Second load criteria is the mass moment of inertia. In correspondence to aforemen-
tioned robot specication (KUKA 6), it must not exceed  kgm
in any case and
shall remain within given constraints of the payload diagram (Figure .) with distances
to the load centre of gravity with respect to the mounting ange on robot axis 6. Thus, it
is necessar y to determine the centre of gravity (CG) and subsequently the mass moment
of inertia (I) following the general formulae
CG
ξ
=
n
i=1
m
i
·r
ξ ,i
n
i=1
m
i
CG
ζ
=
n
i=1
m
i
·r
ζ ,i
n
i=1
m
i
(Eq. .)
I
zz
=
n
i=1
m
i
·r
2
z,i
I = m ·
x
2
+ y
2
+ z
2
(Eq. .)
considering the MWE endeffector draft in Figure .. CG calculation yields
CG
x
= 40.27 ul
CG
z
= 121.33 ul (Eq. .)
in an arbitrarily chosen axis system ξ , ζ marked in Figure .. Translation into robot axis
system x, z by simple zero point offset into mounting ange on robot axis 6 gives actual
values (assuming no CG translation in y-direction)
CG
x
= 40.27 mm CG
z
= 786.33 mm I = 106.71 kg m
2
(Eq. .6)
. ul: unit length
:::
9 Continuous Welding Concept Development
Due to much higher process forces cre-
ated on adherend surface compared to
endeffector equipment weight, the result-
ing centre of gravity shifts far below work-
ing plane. Total mass moment of iner-
tia does not exceed maximum  kg m
,
but CG
z
-position lies outside the given per-
mitted envelope according to Figure ..
This calculation shall be repeated after de-
tailed planning is completed and in agree-
ment with the robot manufacturer which
prescribes “[t]he mass inertia must be ver-
ied using KUKA.Load. It is imperative for
the load data to be entered in the robot
controller. (KUKA 6, 8)
Thus, the use of envisaged and sched-
uled KR - PA is still regarded as fea-
sible.
ξ
ζ
x
z
Figure . Minimum Working Example Draft
. Ultrasonic Welding Equipment
.. Generator
Contrary to static welding, power requirements on the ultrasonic generators are much
higher for continuous applications.
Besides the experimentally used BRANSON :. system ( kHz/ W), the use
case for already existing  kHz systems (:.8) at DLR Augsburg is questioned. Dur-
ing parametric study, mean power ranged between  and 6 W (Table 8.), which
sets the baseline for generator dimensioning. Regarding technical data of both systems
(Table .), it stands evidently to reason that no  kHz-system can provide a sufcient
high continuous power; even lower  kHz-systems are discarded. The already at the
laboratory utilised BRANSON :. system ( kHz/ W) is the most suitable one
for further planning. The use of  kHz-systems in comparable studies (Villegas et al.
; Villegas ; Senders 6) is further justied.
::2
9 Continuous Welding Concept Development
Table . Technical Data of Available Ultrasonic Generators
DCX A/F Model :. :. :. :. :.8
Frequency kHz     
Peak Output Power W     8
Max. Continuous Power W 6    
Data: BRANSON 
.. Sonotrode
Due to abrasive carbon reinforcement of thermoplastic composites and prevailing so-
notrode motion under load, a steel sonotrode is recommended unlike aluminium or tita-
nium sonotrodes show lower wear resistance.
Main challenge is detected as planarity of the sonotrode. Reducing sonotrode dimen-
sions to at least overlap size as well as potential use of a rounded sonotrode are reason-
able countermeasures. For spot welding, Rozenberg found an invariant specic contact
pressure for at unlike a continuous growth of weld areas for spherical sonotrode tips
(, -6) which might be transferable for a more uniform oscillation introduction.
The use of circular sonotrodes might comes along with other melt ow initiation and
guiding over the overlap prole. Yet, such a sonotrode shape was not subject of the
preceding investigations and thus no profound knowledge on such an alternative can
be provided. Variations in sonotrode shape need further investigation.
.6 Energy Directors
The conducted study proved feasibility of at energy directors which are seen as best
suited for a continuous process. Deployment of a neat resin lm can be easily auto-
mated even in actual welding endeffector. Due to the planarity issue, the question of
different ED shapes arose again.
As a matter of fact, shaped EDs provide better focal points particularly for initial melt-
ing and are in turn starting point for melt front propagation (Figure .6a). The most chal-
lenging point is to mould shaped EDs on composite laminates since bres must not be
distorted or destroyed. For hot press manufacturing, Villegas and Palardy () intro-
duced a pre-moulding process with triangular ED strips (Figure .6b). With their height
of . to . mm, they mitigate inaccuracies in planarity remarkably better. However,
for a tape laying process, such a technique is rather complicated to implement. Proled
::3
9 Continuous Welding Concept Development
heated rollers could apply a shaped resin lm as nal step. Still, thickness must be large
enough so xation force does not squeeze pre-moulded EDs back to a at resin layer
without preferential melting properties.
(a) Initial Stage of Molten EDs (b) Schematic for Triangular ED Moulding
Figure .6 Triangular Energy Director Investigation (Villegas and Palardy )
Another technique which is at the forefront of technological revolution is again addi-
tive manufacturing. Besides the opportunity to “print” EDs on the top layer of the lami-
nate, it provides a high degree of freedom for the chosen shape, pattern and thickness.
So far, at energy directors seem to be best choice for an automated process, but if
planarity issue shows unbearable inuence, there are existing techniques taking their
place.
::4
 Conclusion
. Summary
At the beginning, main drivers for recent increase in composite aircraft structure were
shown since there is still high (lightweight) potential compared to the state of the art
“black metal” approach. A detailed overview presented already existing bre-fair fusion
bonding techniques waiting to be implemented. Despite, ultrasonic, resistance and in-
duction welding approaches are the most promising ones for aviation applications. Eval-
uation revealed partly severe disadvantages of one or another fusion bonding process.
Thereby, ultrasonic welding possessed best behaviour and was chosen for further in-
vestigation. Manual pre-testing showed encouraging results for subsequent theoretical
elaborations. Important process parameters were identied and quantied during fol-
lowing parametric study. Moreover, experiments gave a unique insight in the process
and revealed once more its peculiarities and effects. Still, expected parameter corre-
lations were proven in the results discussion. Eventually, this collected knowledge laid
the foundation for continuous process development.
Evidently, ultrasonic welding was chosen due to its persuasive advantages compared
to the other two approaches not due to absence of disadvantages. The rather sim-
ple, quick and clean set-up convinces as well as easy accessibility, ideal continuous
properties and good automation without critical and costly foreign materials. There are
certain drawbacks mainly in form of required planarity and deployment of at energy
directors which need to be solved in order to implement this process successfully in
aircraft series production. Achieved results in front of the background of limited time
show impressively the potential of this technique. Material combinations like CF/PPS
or CF/PEI are investigated intensively these days whereas CF/PEEK still exhibits a lack
of knowledge and experience. Based on this work, there is the chance to change this.
Main challenge for the future will be to establish a stable process incorporating con-
stant joint quality, i.e. particularly areal welds of equal size, controlled melt front propa-
gation and quality monitoring. The transfer from static to continuous will be a challeng-
ing task but worth going onward in this direction. The future in continuous joining will
be ultrasonic.
:0 Conclusion
. Outlook
So far, the main scope contained joining of longitudinal geometries. Nevertheless, this
process can be tailor for circumferential joining of aircraft shell, too (Figure ??). The
small curvature as well as the rather small sonotrode dimensions are fullled conditions
for an three dimensional application. Self-evidently, process control must be stable and
sensor technology must be deployed in a way to enable curved robot paths and main-
tain required process boundaries. Thereby, ultrasonic welding exhibits once more its
versatile and exible nature.
Many approaches go in the direction of not just establishing a stable ultrasonic pro-
cess but develop an in-situ monitoring of key process parameters to track and adjust
them for optimum weld quality. This work showed strongly the various variable inu-
encing and/or disturbing the process. Online measurement and immediate reaction
towards a closed-loop process are the next steps in the direction of automation and in-
dustrialisation of ultrasonic welding particularly in highly regulated aviation industry.
Often referred is “[...] power and displacement data provided for microprocessor con-
trolled ultrasonic welders [that] can be utilised for in situ monitoring of the welding pro-
cess and ultimately of the quality of the welds. (Villegas ) Other observations even
suggest “... that if the output power is be constant, the weld are also will be homoge-
neous. (Senders 6, 6) Whether this hypothesis is proven valid, this will be an out-
standing leverage point for future process controlling.
This approach eventually lays the ground for a sophisticated quality management
collecting data in real time, assessing and recording it for each and every part provid-
ing full traceability as a major requirement for certication in aircraft manufacturing.
Steps are chronologically development and deployment of suitable measurement
technologies for collecting data, analysing for achieving an insight in the process mech-
anisms nally leading to a bundle of parameters setting an envelope for quality control.
Important is the screening of the entire process and extracting of valuable information
(Figure .). A consequent and holistic strategy will provide all tools for a successful
implementation in series production in the future.
::6
:0 Conclusion
Quality
Weld
Fixation
Pneum. Cyl.
Pressure
Stroke
Load Cell
Force
Displacem.
Roller
Radius
Width
Material
Position
Number
Ultrasonic
Lin. Unit
Input Voltage
Input Current
Stroke
Load Cell
Force
Displacem.
Generator
Input Voltage
Input Current
Converter
Input Voltage
Input Current
Amplitude
Frequency
Temperature
Booster
Gain
Material
Sonotrode
Amplitude
Collapse
Material
Shape
Dimensions
Consolidation
Pneum. Cyl.
Pressure
Stroke
Load Cell
Force
Displacem.
Roller
Radius
Width
Material
Position
Number
Figure . Quality Management Schematic
::7
Acknowledgements
Many thanks to the DLR Augsburg for opening my eyes to the world of scientic re-
search, giving me the opportunity to dream, explore, discover and create solely with the
power of mind. Namely my supervisor, Dr. Stefan Jarka who gave me both, a free hand in
developing the idea and concept of the work as well as his expertly competence, knowl-
edge and advice whenever I needed it. The same applies to my academic supervisor
Prof. André Baeten. The way how he suppor ts and demands is inspiring, encouraging
and awakens one’s ambitions to think out of the box and break new ground.
Further thanks to Manuel Endraß and Simon Bauer at DLR Augsburg/Stuttgart for their
words of advice and active suppor t throughout these six months at their institutes.
Big thanks to BRANSON Ultrasonics, namely Daniel Lorenz and Serdar Genc, for their
quick, competent and straightforward activities in preparation and execution of our ex-
perimental study. Their spontaneity and openness to our requests and wishes as well
as their hospitality was commendable, providing a pleasant atmosphere necessary for
professional laboratory work and achieving results of scientic value.
Finally, the greatest and most important thanks go particularly to my parents and fur-
ther my whole family. This work marks the end of my ve year lasting study during which
they supported me extraordinarily, unconditionally and tolerated many hours, days and
weeks of academic work uncomplainingly. Without their love, consolation and encour-
agement I could not evolve into the one I am today. Important achievements and mile-
stones of my life would not be possible. All merits and appreciations belong to them as
well. Unendlichen Dank!
A Experimental Set-Up Addendum
A. Computed Maximum Overlap Length
ASTM D  requires a maximum permissible overlap length as
L
max
=
σ
y
·t
τ
max
(Eq. A.)
with σ
y
as yield strength (for composites: tensile strength), t as thickness of the material,
τ
max
as  % of the estimated average shear strength in joint. For utilised CF/PEEK
plates, data sheets gives a shear strength of  MPa yielding
L
max
=
600 MPa ·2.21 mm
1.5 ·53 MPa
= 16.68 mm > 12.7 mm X (Eq. A.)
fullling the standard requirement.
A Experimental Set-Up Addendum
A. Laminate/Film Properties
Table A. Hauer CF/PEEK Plate Properties
Fibre Torayca T
HT-Carbon, K
Fibre orientation °/°
Weave HS (Harness Satin)
Matrix Victrex G (PEEK)
Fibre Volume Content % ca. 
Density g/cm
ca. .
Tensile Strength MPa 6
Shear Strength MPa 
Young’s Modulus (E
k
) GPa 6.
Young’s Modulus (E
) GPa .6
Poisson Ratio - .
Compressive Strength MPa 
CTE °/° /K x E-6
Max. Service Temperature °C 6
Data: Hauer 
Table A. TORAYCA
®
T Carbon Fibre Properties
Property Test Method Units T
Tensile Modulus TY-B- GPa 
Tensile Strength TY-B- MPa 
Tensile Elongation TY-B- % .
Density TY-B- g/cm
.6
Filament Diameter µm
Thermal Conductivity W / m · K .6
CTE 
-6
/ K -.
Volume Resistivity Ω· cm .
Data: TORAY 
:20
A Experimental Set-Up Addendum
Table A. Aptiv
®
 black PEEK Film Properties
Property Test Method Test Condition Units -G
Glass Transition Temperature ISO  °C 
Melting Point ISO  °C 
Tensile Modulus ISO  °C GPa .
Tensile Strength ISO  (at break) °C MPa 
Tensile Elongation ISO  (at break) °C % >
Shrinkage TM-VX-8 °C %
Density ISO 8 °C g/cm
.
Thermal Conductivity ISO - Average, °C W / m · K .
CTE ISO  Average below T
g

-6
/ K 
Dielectric Strength ASTM D °C kV/mm 
Volume Resistivity ASTM D °C, V Ω· cm .E+6
Data: Victrex 
Table A. LITE
®
TK PEEK Film Properties
Property Test Method Test Condition Units TK
Glass Transition Temperature ISO  °C 
Melting Point ISO  °C 
Tensile Modulus ISO  °C GPa .
Tensile Strength ISO  (at break) °C MPa 
Tensile Elongation ISO  (at break) °C % 
Density ISO 8 °C g/cm
.
Thermal Conductivity DIN 6 W / m · K .
CTE E8 
-6
/ K 6
Dielectric Strength IEC  °C kV/mm 8
Volume Resistivity IEC  Ω· cm .E+8
Data: LITE 
:2:
B Statistical Evaluation
A test result of an arbitrarily chosen measurement series p with the main parameter x
and a limited number of measures must be provided in the form of
p: (x ±x) unit with x =
t ·s
x
n
(Eq. B.)
with standard deviation s
x
of the arithmetic mean value according to
s
x
=
s
1
n 1
n
i=1
(x
i
x)
2
(Eq. B.)
The value for t also known as S’ t-distribution can be taken out of table
works referring to respective probability levels P and considering the number of mea-
sured values n just as given in Table B.. A classication of precision is made, too.
Despite the level of reliability, often times values are referred to the level of failure α, i.e.
P = (1 α) ·100 %.
In this work, nine out of ten specimen provided valid test results after welding and
tension test. Selecting a level of reliability of  %, the related Student’s factor can be
read as 2.262, inserted in (Eq. B.) yields
x =
2.262 ·s
x
9
= 0.754 ·s
x
(Eq. B.)
B Statistical Evaluation
Table B. Student’s t-Distribution Values
Level of Reliability/Failure
orientation operational precision
Number of
Measured
Values
I
σ .6 σ .8 σ σ
P
68. %  %  % . %
α
. . . .
.8 .6 6.6 66.6
. . . .6
. .8 .8 .
. .6 .6 8.6
. . . 6.86
6 . . .
.8 .6 . .8
8 .6 . .
.6 . .8
 .6 .8 .6 .8
 . . . .
 . .86 .8 .8
. .6 .6 .
Source: Kuchling , 6; Zeidler et al. , ; Parthier 6, 
:23
C Fracture Surface Microscopic Analysis
Lower Adherend Upper
#
#
#
#
C Fracture Surface Microscopic Analysis
#
#
#
#6
#8
:25
D Clamping Force Estimation
Clamping screws during experiments are tightened with a torque of  Nm. The rule of
thumb for fastening screws (Wittel et al. , Eq. 8.8) yields
F
f ix
=
M
A
0.17 ·d
=
20 Nm
0.17 ·8 ·10
3
m
= 14705.88 N. (Eq. D.)
This clamping force acts on each screw left and right, thus the resulting reaction force
of specimen is double the xation force. This force is applied by the clamping jaws
with a width of  mm covering the whole specimen width of . mm, i.e. an effective
contact pressure of
p
f ix
=
F
f ix
w
clamp
·w
specimen
=
29411.76 N
20 mm ·25.4 mm
= 57.90 MPa (Eq. D.)
For consolidation and xation, utilisation of rollers is envisaged. In this case, contact
pressure is computed as Hertzian pressure according to (Eq. .)
p
H
=
s
F
N
·E
2πρ ·l
(Eq. D.)
whereof
ρ =
ρ
1
·ρ
2
ρ
1
+ ρ
2
and E =
2 ·E
1
·E
2
(1 ν
2
1
) ·E
2
+ (1 ν
2
2
) ·E
1
(Eq. D.)
with curvature radii ρ
1
, ρ
2
, Young’s moduli E
1
, E
2
, Poissons ratii ν
1
, ν
2
and contact length l
here representing the roller width, i.e. overlap width. For a cylinder-plane contact, ρ = r
roll
and with composite properties (Appendix A) as well as a steel roller, parameters are de-
termined as
E =
2 ·E
roll
·E
(1 ν
2
roll
) ·E
+ (1 ν
2
) ·E
roll
=
2 ·210 GPa ·55.6 GPa
(1 0.3
2
) ·55.6 GPa + (1 0.29
2
) ·210 GPa
= 96.12 GPa
(Eq. D.)
D Clamping Force Estimation
After simplifying, rearranging and inserting, (Eq. D.) reads
F
N
=
p
2
H
·2πr
roll
·l
E
=
(57.90 MPa)
2
·2πr
roll
·l
96.12 ·10
3
MPa
= 0.219 ·r ·l. (Eq. D.6)
Figure D. shows the plot of the function for required force depending on different roller
radii and overlap (roller) width. For a desired overlap of  mm, roller radius must not
exceed about  mm to keep the resulting force below  N.
In order to minimise the distance between the sonotrode and consolidation roller to
apply force as soon as possible to the heated section, rollers should not get too large.
In previous works, rollers exhibited radii of maximum  mm. Corresponding force is
about  N and shall be used for further calculations.
Figure D. Required Force Dependency on Roller Radius and Width with Virtual  N Limit
:27
About the Author
A S is mas ter student for Lightweight Construction and
Composite Technology at the University of Applied Sciences Augsburg. Priorly,
he graduated at the same institution as Bachelor of Engineering in Mechanical
resp. Aerospace Engineering. Within this time, he studied one semester abroad at
the University of Limerick, Ireland, at the Faculty of Aerospace Engineering. Due
to his dual study at Premium AEROTEC GmbH, Augsburg, he already gained
working experience as a manufacturing engineer introduced by his bachelor thesis
on NC Runtime Optimisation of a Clip-Frame-Riveting Robot in the A350 Produc-
tion via Nominal-Actual Feed Analysis (20:6). Besides several company-internal
projects, he was in the lead of the university-approved ALARIS” project Design,
Dimensioning and Production Planning of a Body Wing Aircraft of Composite
Materials (20:5) and the master projects Concept Development of an Integrated Air
Extraction for CFRP Milling Applications (20:6) and Biocomposites in the Field of
Mechanical Engineering (20:7). He lives in Munich.
Bibliography
Ageorges, Christophe, and Lin Ye. a. “Resistance welding of thermosetting composite/ther-
moplastic composite joints. Composites Part A: Applied Science and Manufacturing 
(): 6–6. : -8X. doi:http://doi. org/10.1016/S1359-835X(00)00183-
4.
. b. “Simulation of impulse resistance welding for thermoplastic matrix composites,
8:–. .
. . Fusion Bonding of Polymer Composites. Engineering Materials and Processes.
London: Springer. : 8.
Ageorges, Christophe, Lin Ye, and Meng Hou. a. “Experimental investigation of the resis-
tance welding for thermoplastic-matrix composites: Part I: heating element and heat trans-
fer. Composites Science and Technology 6 (): –. : 66-8. doi:http://
doi.org/10.1016/S0266-3538(00)00005-1.
. b. “Experimental investigation of the resistance welding of thermoplastic-matrix
composites: Part II: optimum processing window and mechanical performance. Compos-
ites Science and Technology 6 (8): –. : 66-8. doi:http://doi.org/10.
1016/S0266-3538(00)00025-7.
. . “Advances in fusion bonding techniques for joining thermoplastic matrix compos-
ites: a review. Composites Part A: Applied Science and Manufacturing  (6): 8–8.
: -8X. doi:http://dx.doi.org/10.1016/S1359-835X(00)00166-4.
AH, (Airbus Helicopters). 6. Internal Meeting/Unpublished.
Ahmed, N. . New Developments in Advanced Welding. Woodhead Publishing Series in Weld-
ing and Other Joining Technologies. Elsevier Science. : 8868.
Ahmed, T.J., D. Stavrov, H.E.N. Bersee, and A. Beukers. 6. “Induction welding of thermoplastic
composites—an overview. Composites Part A: Applied Science and Manufacturing  ():
68–6. : -8X. doi:http://doi.org/10.1016/j.compositesa.2005.10.
009.
Airbus. 6. “Taking the lead: the A XWB. Accessed April , . https://www.airbusgroup.
com / dam / assets / airbusgroup / int / en / investor - relations / documents / 2006 /
untitled/further_pre_a350_xwb_launch_2006.pdf.
AM. . “More Control for Ultrasonic Welding. Assembly Magazine (June).
Arias, M., and G. Ziegmann. 6. “Impulse resistance welding: a new technique for joining ad-
vanced thermoplastic composite parts. In Proceedings of the st International SAMPE
Symposium and Exhibition. Anaheim.
ASTM D , (Deutsches Institut für Normung). . Standard Test Method for Apparent Shear
Strength of Single-Lap-Joint Adhesively Bonded Metal Specimens by Tension Loading (Met-
Bibliography
al-to-Metal). Deutsche Industrienorm. ASTM International (American Society for Testing
and Materials), West Conshohocken. doi:10.1520/D1002-10.
Baehr, Hans Dieter, and Karl Stephan. 6. Wärme- und Stoffübertragung. Berlin Heidelberg:
Springer. : 8666.
Bargel, Hans-Jürgen, and Günter Schulze. . Werkstoffkunde. Springer-Lehrbuch. Berlin and
Heidelberg: Springer. : 86.
Bayerl, Thomas, Miro Duhovic, Peter Mitschang, and Debes Bhattacharyya. . “The heating of
polymer composites by electromagnetic induction A review. Composites Part A: Applied
Science and Manufacturing :–. : -8X. doi:http://doi.org/10.1016/j.
compositesa.2013.10.024.
Beards, C. . Engineering Vibration Analysis with Application to Control Systems. Elsevier Sci-
ence. : 886.
Beevers, A. . “Welding: the way ahead for thermoplastics?” ACE-, Engineering: .
Beland, Sylvie. . High Performance Thermoplastic Resins and Their Composites. Elsevier
Science. : 88.
Benatar, A., J.W. Gillespie Jr., and K. Kedward. . “Joining of polymer composites. Chap. 
in Advanced Composites Manufacturing, by T.G. Gutowski. New York: John Wiley / Sons.
: 86.
Benatar, Avraham, and Zuhang Cheng. 8. “Ultrasonic welding of thermoplastics in the far--
eld. Polymer Engineering & Science  (): 6–. : 8-6. doi:10.1002/
pen.760292312.
Benatar, Avraham, Raman V. Eswaran, and Satinder K. Nayar. 8. “Ultrasonic welding of ther-
moplastics in the near-eld. Polymer Engineering & Science  (): 68–68. : 8-6.
doi:10.1002/pen.760292311.
Benatar, Avraham, and Timothy G. Gutowski. 86. “Methods for fusion bonding thermoplastic
composites. SAMPE Quaterly 8 (): –.
. 88. “Ultrasonic Welding of Advanced Thermoplastic Composites. In rd International
SAMPE Symposium, 8–.
. 8. “Ultrasonic welding of PEEK graphite APC- composites. Polymer Engineering &
Science  (): –. : 8-6. doi:10.1002/pen.760292313.
Benkowsky, G. . Induktionserwärmung: Härten, Glühen, Schmelzen, Löten, Schweissen. Grund-
lagen und praktische Anleitungen für Induktionserwärmungsverfahren, insbesondere auf
dem Gebiet der Hochfrequenzerwärmung. Verlag Technik. : 88.
BMW Group. . “How a BMW i is created: The LifeDrive concept in four steps. Accessed
April , . http://www.bmwgroup- plants.com/en/leipzig/production/bmw- i-
composition.html.
:30
Bibliography
Boeing. 6. “Jährliche Wachstumsraten für den Passagierbereich im Luftverkehr von 6 bis
 nach Regionen (gemessen an Revenue-Passagier-Kilometern). Statista. Accessed
April , . https://de.statista.com/statistik/daten/studie/165532/umfrage/
wachstumsraten-fuer-den-passagier--und-cargobereich-im-luftverkehr/.
Border, J., and R. Salas. 8. In Proceedings of the th International SAMPE Symposium, 6–
8.
Boyard, Nicolas. 6. Heat Transfer in Polymer Composite Materials: Forming Processes. Me-
chanical engineering and solid mechanics series. Wiley. : 8886.
BRANSON. . Ultrasonic Welding Characteristics of Textiles and Films. PW-, Danbury.
. . DCX A and F Power Supplies. PJ--.
. . Xc: Ultraschall-Fertigungssysteme für Kunststoffschweißen. PJ-8-.
Campbell, Flake C. . Manufacturing Processes for Advanced Composites. Elsevier Science.
: 888.
Chursin, Alexander, and Yury Makarov. . Management of Competitiveness: Theory and Prac-
tice. Springer International Publishing. : 86.
Clean Sky JU. 6. Clean Sky at a Glance. Brussels, March.
Cole, K.C. . “A review of Recent Developments in Joining High-Performance Thermoplastic
Composites. In Proceedings of Canadian International Composites Conference and Exhibi-
tion, edited by S.V. Hoa and R. Gauvin, –8.
Costa, Anahi Pereira da, Edson Cocchieri Botelho, Michelle Leali Costa, Nilson Eiji Narita, and
José Ricardo Tarpani. . “A Review of Welding Technologies for Thermoplastic Com-
posites in Aerospace Applications. Journal of Aerospace Technology and Management
(): –6. : . doi:10.5028/jatm.2012.04033912.
CUP, (Cambridge University Press). 8. Cambridge Advanced Learner’s Dictionary. Cambridge
University Press. : 8668.
Davies, P., W.J. Cantwell, P.-Y. Jar, P.-E. Bourban, V. Zysman, and H.H. Kausch. . “Joining
and repair of a carbon bre-reinforced thermoplastic. Composites  (6): –. :
-6. doi:http://dx.doi.org/10.1016/0010-4361(91)90199-Q.
Delgado Labrandero, Sofía. . “Characterization of metallic meshes used for resistance weld-
ing of thermoplastic composites. Ingeniería Industrial Thesis, Universidad Carlos III de
Madrid.
DIN -, (Deutsches Institut für Normung). . Schweißen und verwandte Prozesse - Be-
griffe: Teil : Metallschweißprozesse mit Ergänzungen zu DIN EN 6:. Deutsche
Industrienorm. Deutsches Institut für Normung.
DLR, (Deutsches Zentrum für Luft- und Raumfahrt). . DLR-Luftfahrtforschung. Cologne, De-
cember.
. a. Internal Meeting/Unpublished.
:3:
Bibliography
DLR, (Deutsches Zentrum für Luft- und Raumfahrt). b. “DLR at JEC World  in Paris.
March . Accessed May , . http://www.dlr.de/dlr/en/desktopdefault.aspx/
tabid-10081/151_read-21690/#/gallery/26534.
Domininghaus, Hans. . Kunststoffe: Eigenschaften und Anwendungen. Edited by Peter El-
sner, Peter Eyerer, and Thomas Hirth. VDI-Buch. Berlin-Heidelberg: Springer. : 866.
Don, R.C., L. Bastien, T.B. Jakobsen, and J.W. Gillepsie Jr. . “Fusion bonding of thermoplastic
composites by reistance heating. SAMPE Journal 6:–6.
Dorsch, Jessica Katrin. . “Experimentelle Ermittlung der Prozessparameter zum automa-
tisierten Ultraschall-Heften von Carbonfaser-verstärkten Thermoplast-Lagen. Master The-
sis, University of Applied Sciences Augsburg.
Dubé, Martine. . “Static and Fatigue Behaviour of Thermoplastic Composite Laminates
Joined by Resistance Welding. 86. PhD Thesis, McGill University Montréal.
Dubé, Martine, P. Hubert, A. Yousefpour, and J. Denault. 8a. “Current leakage prevention in re-
sistance welding of carbon bre reinforced thermoplastics. Composites Science and Tech-
nology 68:–8. : 66-. doi:10.1016/j.compscitech.2007.09.008.
Dubé, Martine, Pascal Hubert, Jan N.A.H. Gallet, Darko Stavrov, Harald E.N. Bersee, and Ali Yousef-
pour. 8b. “Fatigue performance characterisation of resistance-welded thermoplastic
composites. Composites Science and Technology 68 (–8): –6. : 66-8.
doi:http://doi.org/10.1016/j.compscitech.2008.02.012.
Dubé, Martine, Pascal Hubert, Ali Yousefpour, and J. Denault. . “Fatigue failure character-
isation of resistance-welded thermoplastic composites skin/stringer joints. International
Journal of Fatigue  (): –. : -. doi:http : //doi .org/ 10.1016 /j.
ijfatigue.2008.03.012.
Dubé, Martine, Pascal Hubert, Ali Yousefpour, Johanne Denault, and Matthew Wadham-Gagnon.
. “Resistance welding of Thermoplastic Composite Skin/Stringer Specimens. Com-
posites Part A Applied Science and Manufacturing 8, no.  (December). doi:10.1016/j.
compositesa.2007.07.014.
Duhovic, Miro, Joachim Hausmann, Pierre L’Eplattenier, and Inaki Caldichoury. . “A Finite
element Investigation into the Continuous Induction Welding of Dissimilar Material Joints.
In Proceedings of the th European LS-DYNA Conference .
Dukane. . Guide to Ultrasonic Plastics Assembly. -6-. St. Charles: Dukane Intelligent
Assembly Solutions, August.
EC, (European Commission). . “End-of-life aircraft recycling offers high grade materials.
June . Accessed May , . http://ec.europa.eu/environment/ecoap/about-
eco-innovation/good-practices/eu/719_en.
Ehrenstein, Gottfried W., Gabriela Riedel, and Pia Trawiel. . Thermal Analysis of Plastics: The-
ory and Practice. München: Carl Hanser Verlag GmbH & Company KG. : 86.
EU, (European Union). . Directive on Landll of Waste. //EC, April 6.
:32
Bibliography
EU, (European Union). . Directive on End-of Life Vehicle. //EC, September 8.
Eveno, C.E., and J.W. Gillespie Jr. 8. “Experimental Investiagtion of Ultrasonic Welding of
Graphite Reinforced Polyetheretherketone Composites. st International SAMPE Techni-
cal Conference, no. : –.
Eveno, E.C., and J.W. Gillespie Jr. 88. “Resistance welding of graphite polyetheretherketone
composites: an experimental investigation. Journal of Thermoplastic Composite Materials
:–8.
FAA, (Federal Aviaton Administration). . Stress Analysis of In-Plane, Shear-Loaded, Adhe-
sively Bonded Composite Joints and Assemblies. Final Report DOT/FAA/AR-/. U.S. De-
partment of Transportation, April.
. . Aviation Maintenance Technician Handbook: Airframe. Vol. . Newcaslte: Aviation
Supplies & Academics Inc. : 86.
Fernie, J.A., P.L. Threadgill, and M.N. Watson. . “Progress in Joining Advanced Materials.
Welding and Metal Fabrication (June): –8.
Fischer, Frederic, Yannick Mezakeu Tongnan, Matthias Beyrle, Tobias Gerngross, and Michael
Kupke. . “Characterization of Production-Induced Defects in Carbon Fiber Reinforced
Thermoplastic Technology. Edited by Randolf Hanke, Christian Boller, and Matthias Purschke.
Bremen, November.
Fish, J.C., M.L. Vitlip, and S.P. Chen. . “Interlaminar Fracture Characteristics of Bonding Con-
cepts for Thermoplastic Primary Structures. AIAA Journal :6–68.
Fokker. . New evolution: Progress in thermoplastic composites. Brochure. Papendrecht: Fokker
Aerostructures.
Frantz, J. . “Joining Plastics the Sound Way. Machine Design, no. : 6–6.
Freist, Carsten. . “Experimentelle und numerische Untersuchungen zum Widerstandsschweißen
endlosfaser- und kurzfaserverstärkter thermoplastischer Hochleistungsstrukturen. Disser-
tation, Universität Stuttgart.
Gachnang, Hans Rudolf. 8. Ultrasonic welding device. EP Patent App. EP,6,,,
May.
Gallego-Juárez, J.A., and K.F. Graff. . Power Ultrasonics: Applications of High-Intensity Ultra-
sound. Woodhead Publishing Series in Electronic and Optical Materials. Elsevier Science.
: 886.
Gardiner, Ginger. . “Thermoplastic composites: Primary structure?” February . Accessed
May , . http://www.compositesworld.com/articles/thermoplastic-composites-
primary-structure.
Gerdeen, James C., James C. Gerdeen, Harold W. Lord, and Ronald A.L. Rorrer. . Engineering
Design with Polymers and Composites. Engineering Design with Polymers and Composites,
Vol. . Boca Raton: Taylor & Francis. : 888.
:33
Bibliography
Gleich, D.M., M.J.L van Tooren, and A. Beukers. . “Analysis and evaluation of bondline thick-
ness effects on failure load in adherends bonded structures. Journal of Adhesion Science
and Technology  (): –.
Grewell, D.A., A. Benatar, and J.B. Park. . “Ultrasonic welding. Chap. 8 in Plastics and Com-
posites Welding Handbook. Munich: Hanser. : 6.
Grimm, R.A. . “Fusion Welding Techniques. Welding Journal 6:–8.
Guess, T.R., and R.E. Allred. . “Comparison of Lap Shear Specimens. Journal of Testing and
Evaluation, no. : 8–.
Haimbaugh, R.E. . Practical Induction Heat Treating. ASM International. : 868.
Harras, B., K. C. Cole, and T. Vu-Khanh. 6. “Optimization of the Ultrasonic Welding of PEEK--
Carbon Composites. Journal of Reinforced Plastics and Composites  (): –8. doi:10.
1177/073168449601500203. eprint: http://dx.doi.org/10.1177/073168449601500203.
Hauer, (Hauer Composites GmbH & Co. KG). . Carbon PEEK Platten. Datasheet. Blaubeuren.
Heimerdinger, M.W. . “Repair Technology for Thermoplastic Aircraft Structures. Composite
Repair of Military Aircraft Structures: th Meeting of the AGARD Structures and Materials
Panel, Seville (Qudbec), no. AGARD-CP- (January): –.
Ho, Janet, and Richard Jow. . Characterization of High Temperature Polymer Thin Films for
Power Conditioning Capacitors. Report. ARL-TR-88. Adelphi: Army Research Laboratory,
July.
Holmes, S.T., R.C. Don, J.W. Gillespie Jr., I. Howie, and A. Ortona. . “Sequential resistance
welding of large-scale IM/PEEK double lap joints. In CCM Report -. Centre for Com-
posite Materials, University of Delaware.
Hopmann, Christian, and Walter Michaeli. . Einführung in die Kunststoffverarbeitung. München:
Carl Hanser Verlag GmbH & Company KG. : 8668.
Hou, M., and K. Friedrich. . “Resistance welding of continuous carbon bre/polypropylene
composites. Plastic, Rubber and Composites Processing and Applications 8 (): –.
Hou, Meng, and K. Friedrich. . “Resistance welding of continuous glass bre-reinforced
polypropylene composites. Composites Manufacturing (): –6. : 6-.
doi:http://dx.doi.org/10.1016/0956-7143(92)90078-9.
Hou, Meng, Mingbo Yang, Andrew Beehag, Yiu-Wing Mai, and Lin Ye. a. “Resistance weld-
ing of carbon bre reinforced thermoplastic composite using alternative heating element.
Tenth International Conference on Composite Structures, Composite Structures  (–):
66–6. : 6-8. doi:http://dx.doi.org/10.1016/S0263-8223(00)00047-7.
Hou, Meng, L. Ye, and Y.-W. Mai. b. “An Experimental Study of Resistance Welding of Car-
bon Fibre Fabric Reinforced Polyetherimide (CF Fabric/PEI) Composite Material. Applied
Composite Materials 6 (): –. : -8. doi:10.1023/A:1008879402267.
:34
Bibliography
Hussey, Robert John, and Josephine Wilson. . Advanced Technical Ceramics Directory and
Databook. Springer US. : 8866.
Irshad, K.T. . “Anvil Design & Experimental Investigation for Ultrasonic Welding of Thin Dis-
similar Metals. Master Thesis, National Institute of Technology Rourkela.
Irving, E.P., and C. Soutis. . Polymer Composites in the Aerospace Industry. Woodhead Pub-
lishing series in composites science and engineering. Cambridge: Elsevier Science. :
888.
Jakobsen, Tom B., Roderic C. Don, and John W. Gillespie. 8. “Two-Dimensional thermal anal-
ysis of resistance welded thermoplastic composites. Polymer Engineering & Science 
(): –. : 8-6. doi:10.1002/pen.760292314.
Jen, Ming-Hwa R., Yu-Chung Tseng, Huang-Kuang Kung, and J.C. Huang. 8. “Fatigue re-
sponse of APC- composite laminates at elevated temperatures. Composites Part B: En-
gineering  (–8): –6. : -868. doi:http : / / doi . org / 10 . 1016 / j .
compositesb.2008.03.005.
Jiang, Han, Robert Browning, Jason Fincher, Anthony Gasbarro, Scooter Jones, and Hung-Jue
Sue. 8. “Inuence of surface roughness and contact load on friction coefcient and
scratch behavior of thermoplastic olens. Applied Surface Science  (): –.
: 6-. doi:http://dx.doi.org/10.1016/j.apsusc.2008.01.067.
Jouin, P., T. Lee, and R. Vitlip. . “Manufacture of a Primary Flight Structure using Thermo-
plastics. Proceedings of the 6th International SAMPE Symposium: –.
Kaczmar, J.W., K. Pietrzak, and W. Włosiński. . “The production and application of metal
matrix composite materials. Journal of Materials Processing Technology 6 (–): 8–
6. : -6. doi:http://dx.doi.org/10.1016/S0924-0136(00)00639-7.
Kagan, Val A., and Russell J. Nichols. . “Benets of Induction Welding of Reinforced Thermo-
plastics in High Performance Applications. Journal of Reinforced Plastics and Composites
 (): –. doi:10. 1177/0731684405048846. eprint: http:/ /dx.doi.org/10 .
1177/0731684405048846.
Kaiser, Wolfgang. . Kunststoffchemie für Ingenieure: Von der Synthese bis zur Anwendung.
München: Carl Hanser Verlag GmbH & Company KG. : 86.
Khmelev, V.N., A.N. Slivin, and A.D. Abramov. . “Model of process and calculation of energy
for a heat generation of a welded joint at ultrasonic welding polymeric thermoplastic mate-
rials. In Electron Devices and Materials, . EDM’. 8th Siberian Russian Workshop and
Tutorial on, 6–. IEEE.
Khmelev, V.N., A.N. Slivin, A.V. Lehr, and A.D. Abramov. . “Theoretical investigations of con-
tinuous ultrasonic seam welding of thermoplastic polymers and fabrics. In  th Inter-
national Conference and Seminar on Micro/Nanotechnologies and Electron Devices, –
. June. doi:10.1109/EDM.2010.5568791.
:35
Bibliography
Krüger, S., G. Wagner, and D. Eier. . “Ultrasonic Welding of Metal/Composite Joints. Ad-
vanced Engineering Materials 6 (): –. : -68. doi:10.1002/adem.200300539.
Ku, H.S. . “Joining of Thirty Three Percent by Weight Random Carbon Fibre Reinforced Low
Density Polyethylene using Variable Frequency Microwave, University of Southern Queens-
land.
Kuchling, Horst. . Taschenbuch der Physik. Leipzig: Hanser Fachbuchverlag. : 86.
KUKA, (KUKA Roboter GmbH). 6. Robots KR - PA, KR - PA. Spez KR  - PA.
Specication. Augsburg, March 6.
Labeas, G.N., G.A. Moraitis, and Ch.V. Katsiropoulos. . “Optimization of Laser Transmis-
sion Welding Process for Thermoplastic Composite Parts using Thermo-Mechanical Sim-
ulation. Journal of Composite Materials  (): –. : -8. doi:10 .1177/
0021998309345325.
Lambing, C.L.T., S.M. Andersen, S. Holmes, R. Don, S. Leach, and J. Gillespie. . Apparatus
and method for resistance welding. US Patent ,,, July.
Lambing, C.L.T., R.C. Don, S.M. Andersen, S.T. Holmes, B.S. Leach, and J.W. Gillespie Jr. .
“Design and Manufacture of an Automated Resistance Welder for Thermoplastic Compos-
ites. In Proceedings of the th Annual Technical Conference, –.
Leatherman, A.F. . Composite plastic-metallic bonding means and method. US Patent App.
/,, June.
Levy, Arthur, Steven Le Corre, and Irene Fernandez Villegas. a. “Modeling of the heating
phenomena in ultrasonic welding of thermoplastic composites with at energy directors.
Journal of Materials Processing Technology  (): 6–. : -6. doi:http:
//dx.doi.org/10.1016/j.jmatprotec.2014.02.009.
Levy, Arthur, Steven Le Corre, and Arnaud Poitou. b. “Ultrasonic welding of thermoplastic
composites: a numerical analysis at the mesoscopic scale relating processing parameters,
ow of polymer and quality of adhesion. International Journal of Material Forming (): –
. : 6-6. doi:10.1007/s12289-012-1107-6.
Lewis, C.F. . Materials Science and Engineering , no. 6 (June): –.
Li, Hui, and Karl Englund. . “Recycling of carbon ber-reinforced thermoplastic composite
wastes from the aerospace industry. Journal of Composite Materials  (): 6–.
doi:10.1177/0021998316671796. eprint: http://dx.doi.org/10.1177/0021998316671796.
Lin, W., O. Buneman, and A.K. Miller. . “Induction heating model for graphite ber/thermo-
plastic matrix composites. Sampe Journal :–.
LITE, (LITE GmbH). . LITE TK (PEEK). Datasheet. Gaenz.
Liu, G., and A.N. Gent. . A triangular double cantilever beam test for measuring adhesive or
cohesive fracture energy. Technical Report . AD-A /N-8-K-/-.
Akron: Institute of Polymer Engineering, University of Akron.
:36
Bibliography
Löwer, Chris. . “Ultraleicht und extrem stabil. weiter.vorn, no. : 6. : 868-8.
Lu, H.M., A. Benatar, and F.G. He. . “Sequential ultrasonic welding of peek/Graphite compos-
ite plates. In Society of Plastics Engineers Annual Technical Conference, –6.
Luratec. . “CFK Next Generation: Luftfahrttechnik. Accessed April , . http://www.
luratec.com/11fo/01lt/cf01.jpg.
Maguire, D.M. 8. “Joining Thermoplastic Composites. SAMPE Journal  (): –.
Mahon, J., C. Rutkowski, and W. Oelcher. . “Induction bonding procedures for graphite rein-
forced thermoplastic assemblies. In Proceedings of the rd International SAMPE Techni-
cal Conference, –8. A- -. Covina: Society for the Advancement of Mate-
rial / Process Engineering.
Markatos, D.N., K.I. Tserpes, E. Rau, S. Markus, B. Ehrhar t, and Sp. Pantelakis. . “The ef-
fects of manufacturing-induced and in-service related bonding quality reduction on the
mode-I fracture toughness of composite bonded joints for aeronautical use. Composites
Part B: Engineering  (): 6–6. : -868. doi:https://doi.org/10.1016/j.
compositesb.2012.05.052.
Martienssen, Werner, and Hans Warlimont. 6. Springer Handbook of Condensed Matter and
Materials Data. Springer Handbook of Condensed Matter and Materials Data. Berlin and
Heidelberg: Springer. : 8.
Matsen, M.R. . Combined inductive heating cycle for sequential forming the brazing. US
Patent ,,, May.
. . Superplastically formed part. US Patent ,,, December.
Matsen, M.R., and A.R. Hodges. 8. Internal tooling for induction heating. US Patent ,,,
February.
Matsen, M.R., and D.A. McCarville. 8. Bonding using induction heating. US Patent ,,,
August.
Matsen, M.R., R.J. Poel, W.B. Crow, and D.S. Nansen. 6. Joining large structures using local-
ized induction heating. WO Patent App. PCT/US6/,6, December.
Mattia, F.B. . Elsevier’s Dictionary of Acronyms, Initialisms, Abbreviations and Symbols. El-
sevier Science. : 88.
McCarville, Douglas A., and Henry A. Schaefer. . Processing and Joining of Thermoplastic
Composites. ASM Handbook. ASM International. : 8.
McKnight, Steven H., Scott T. Holmes, John W. Gillespie, Cynthia L. T. Lambing, and James M.
Marinelli. . “Scaling issues in resistance-welded thermoplastic composite joints. Ad-
vances in Polymer Technology 6 (): –. : 8-. doi:10.1002/(SICI)1098-
2329(199711)16:4<279::AID-ADV3>3.0.CO;2-S.
:37
Bibliography
Menges, Georg, Edmund Haberstroh, Walter Michaeli, and Ernst Schmachtenberg. . Menges
Werkstoffkunde Kunststoffe. München: Carl Hanser Verlag GmbH & Company KG. :
86.
Metaxas, A.C., and Roger J. Meredith. 8. Industrial Microwave Heating. Chap. . Energy Engi-
neering Series, Bd. . P. Peregrinus. : 8688.
Mitchell, M.R., and R.W. Landgraf. 6. Advances in Fatigue Lifetime Predictive Techniques.
ASTM special technical publication, Bd. . ASTM. : 88.
Mitschang, P., R. Rudolf, and M. Neitzel. . “Continuous Induction Welding Process, Modelling
and Realisation. Journal of Thermoplastic Composite Materials  (): –. doi:10.
1177/0892705702015002451. eprint: http://dx.doi.org/10.1177/0892705702015002451.
Moser, Lars, Peter Mitschang, and Alois K. Schlarb. 8. “Induction welding: an automated
joining process for composites. JEC Magazine  (November): –. : 6-6X.
MTorres. . “R & D Laboratory for Composite Materials. Accessed November , . http:
//www.mtorres.es/en/aeronautics/laboratory.
Neitzel, Manfred, Peter Mitschang, and Ulf Breuer. . Handbuch Verbundwerkstoffe: Werk-
stoffe, Verarbeitung, Anwendung. nd ed. München: Carl Hanser Verlag GmbH & Company
KG. : 8668.
Niu, C., and M.C.Y. Niu. . Airframe Structural Design: Practical Design Information and Data on
Aircraft Structures. Airframe book series. Adaso Adastra Engineering Center. : 868.
Noor, Ahmed Khairy. . Structures Technology for Future Aerospace Systems. Progress in as-
tronautics and aeronautics. American Institute of Aeronautics & Astronautics. : 8686.
NRC, (National Research Council). 6. Accelerated Aging of Materials and Structures: The Ef-
fects of Long-Term Elevated-Temperature Exposure. Edited by Committee on Evaluation of
Long-Term Aging of Materials, Commission on Engineering Structures Using Accelerated
Test Methods, and National Research Council Technical Systems. Washington D.C.: Na-
tional Academies Press. : 866.
NRCC, (National Research Council Canada). . “Resistance welding of thermoplastic com-
posites. Government of Canada. September 6. http : //www. nrc - cnrc .gc. ca / eng/
solutions/facilities/amtc/thermoplastic_composites.html.
Offringa, Arnt. . “New thermoplastic composite design concepts and their automated man-
ufacture. JEC Composites Magazine, no.  (June): –.
O’Shaughnessey, Patrice Gouin, Martine Dubé, and Irene Fernandez Villegas. 6. “Modeling
and experimental investigation of induction welding of thermoplastic composites and com-
parison with other welding processes. Journal of Composite Materials  (): 8–.
doi:10.1177/0021998315614991. eprint: http://dx.doi.org/10.1177/0021998315614991.
Palardy, G., F.M. Agricola dn C. Dransfeld, and I.F. Villegas. . Up-Scaling of the Ultrasonic
Welding Process for Joining Thermoplastic Composites. Presentation, May.
:38
Bibliography
Palardy, Genevieve, and Irene Fernandez Villegas. 6. “Smart Ultrasonic Welding of Thermo-
plastic Composites. In Proceedings of the American Society for Composites: Thirty-First
Technical Conference, edited by B.D. Davidson, M.W. Czabaj, and J.G. Ratcliffe.
. . “On the effect of at energy directors thickness on heat generation during ultra-
sonic welding of thermoplastic composites. Composite Interfaces  (): –. doi:10.
1080/09276440.2016.1199149. eprint: http://dx.doi.org/10.1080/09276440.2016.
1199149.
Pappadà, Silvio, Andrea Salomi, Jeanette Montanaro, Alessandra Passaro, Antonio Caruso, and
Alfonso Maffezzoli. . “Fabrication of a thermoplastic matrix composite stiffened panel
by induction welding. Aerospace Science and Technology :–. : -68.
doi:https://doi.org/10.1016/j.ast.2015.03.013.
Parthier, Rainer. 6. Messtechnik: Grundlagen und Anwendungen der elektrischen Me sstechnik.
Wiesbaden: Springer Fachmedien. : 8688.
PDL, (Plastics Design Library). 8. Handbook of Plastics Joining: A Practical Guide. Edited by
Michael J. Troughton. Plastics Design Library. Elsevier Science. : 88.
Pickering, S.J. 6. “Recycling technologies for thermoset composite materials—current sta-
tus. The nd International Conference: Advanced Polymer Composites for Structural Ap-
plications in Construction, Composites Part A: Applied Science and Manufacturing  (8):
6–. : -8X. doi:https://doi.org/10.1016/j.compositesa.2005.05.
030.
Pimenta, Soraia, and Silvestre T. Pinho. . “Recycling carbon bre reinforced polymers for
structural applications: Technology review and market outlook. Environmental Implica-
tions of Alternative Materials in Construction and Treatment of Waste, Waste Management
 (): 8–. : 6-X. doi:https://doi.org/10.1016/j.wasman.2010.09.
019.
Potente, Helmut. . Fügen von Kunststoffen: Grundlagen, Verfahren, Anwendung. München:
Hanser. : 86.
Puyal, D., C. Bernal, J. M. Burdio, J. Acero, and I. Millan. . “Methods and procedures for accu-
rate induction heating load measurement and characterization. In  IEEE International
Symposium on Industrial Electronics, 8–8. June. doi:10.1109/ISIE.2007.4374700.
Qin, Tianliang, Libin Zhao, and Jianyu Zhang. . “Fastener effects on mechanical behav-
iors of double-lap composite joints. Composite Structures :–. : 6-8.
doi:http://dx.doi.org/10.1016/j.compstruct.2013.01.008.
Rao, N. Shama, T.G.A. Simha, K.P. Rao, and G.V.V. Ravi Kumar. . Carbon Composites Are
Becoming Competitive And Cost Effective. Whitepaper. Infosys.
Rapoport, E., and Y. Pleshivtseva. 6. Optimal Control of Induction Heating Processes. Me-
chanical Engineering. CRC Press. : 8.
:39
Bibliography
Rodgers, B.A., and P.J. Mallon. . In Proceedings of the th SAMPE International European
Chapter Conference, –.
Rose, J.L., F. Yan, Y. Liang, and C. Borigo. . “Ultrasonic Vibration Method for Damage Detec-
tion in Composite Aircraft Components. In Topics in Modal Analysis II, Volume 6: Proceed-
ings of the th IMAC, A Conference on Structural Dynamics, , edited by R. Allemang, J.
De Clerck, C. Niezrecki, and J.R. Blough, –6. New York: Springer. : 86.
doi:10.1007/978-1-4614-2419-2_36.
Rotheiser, Jordan. . Joining of Plastics: Handbook for Designers and Engineers. Hanser Pub-
lishers. : 86.
Rozenberg, L. . Physical Principles of Ultrasonic Technology. Ultrasonic Technology. Springer
US. : 8688.
Rudolf, R., P. Mitschang, and M. Neitzel. . “Induction heating of continuous carbon-bre-re-
inforced thermoplastics. Composites Part A: Applied Science and Manufacturing  ():
–. : -8X. doi:http://doi.org/10.1016/S1359-835X(00)00094-4.
Rudolf, R., P. Mitschang, M. Neitzel, and C. Rueckert. . “Welding of High-Performance Ther-
moplastic Composites. 86, Polymers & Polymer Composites , no. (January):
–.
Sakamoto, Shigeyasu. . Beyond World-Class Productivity: Industrial Engineering Practice and
Theory. Chap. 6. London: Springer. : 88668.
Sanders, Peter. 8. “Electromagnetic welding: an advance in thermoplastics assembly. Ma-
terials & Design 8 (): –. : 6-6. doi:http://dx.doi.org/10.1016/0261-
3069(87)90059-8.
Schinner, G., J. Brandt, and H. Richter. 6. “Recycling Carbon-Fiber-Reinforced Thermoplastic
Composites. Journal of Thermoplastic Composite Materials (): –. doi:10.1177/
089270579600900302. eprint: http://dx.doi.org/10.1177/089270579600900302.
Schulshenko, M.N. . Konstruktion von Flugzeugen. Leipzig: Elbe-Dnjepr-Verlag. : 8.
Schulze, M. . “Produktion des BMW i setzt neue Maßstäbe. VDI Nachrichten, no.  (Oc-
tober ).
Schürmann, Helmut. . Konstruieren mit Faser-Kunststoff-Verbunden. nd ed. VDI-Buch. Berlin
and Heidelberg: Springer-Verlag. : 8.
Schwartz, M.M. . Joining of Composite Materials.
Senders, F.J.M. 6. “Continuous Ultrasonic Welding of Thermoplastic Composites. Master
Thesis, Delft University of Technology.
Senders, Frank, Martijn van Beurden, Genevieve Palardy, and Irene F. Villegas. 6. “Zero-ow:
a novel approach to continuous ultrasonic welding of CF/PPS thermoplastic composite
plates. Advanced Manufacturing: Polymer & Composites Science (-): 8–. doi:10.
1080/20550340.2016.1253968. eprint: http://dx.doi.org/10.1080/20550340.2016.
1253968.
:40
Bibliography
Shehab, Essam, Weitao Ma, and Ahmad Wasim. . “Manufacturing Cost Modelling for Aerospace
Composite Applications. In Concurrent Engineering Approaches for Sustainable Product
Development in a Multi-Disciplinary Environment: Proceedings of the th ISPE International
Conference on Concurrent Engineering, edited by Josip Stjepandić, Georg Rock, and Cees
Bil, –. London: Springer. : 86. doi:10.1007/978- 1- 4471- 4426-
7_37.
Shi, H. . “Resistance Welding of Thermoplastic Composites: Process and Performance.
PhD Thesis, Delft University of Technology. doi:10. 4233 / uuid:58f16f99 - afc1 - 4713 -
842e-562412583340.
Shi, Huajie, Irene Fernandez Villegas, Marc-André Octeau, Harald E.N. Bersee, and Ali Yousefpour.
. “Continuous resistance welding of thermoplastic composites: Modelling of heat gen-
eration and heat transfer. Composites Part A: Applied Science and Manufacturing :6–
6. : -8X. doi:http://doi.org/10.1016/j.compositesa.2014.12.007.
Siddiq, A., and E. Ghassemieh. 8. “Thermomechanical analyses of ultrasonic welding pro-
cess using thermal and acoustic softening effects. Mechanics of Materials  (): 8–
. : 6-666. doi:http://doi.org/10.1016/j.mechmat.2008.06.004.
Silverman, E.M., and B.A. Griese. 8. “Joining methods for graphite/PEEK thermoplastic com-
posites. SAMPE Journal  (): –8.
Siores, E., and D.D. Rego. . “Joining Materials with Microwave Energy. Australasian Welding
Journal: 8–.
Smiley, A. J., A. Halbritter, F. N. Cogswell, and P. J. Meakin. . “Dual polymer bonding of ther-
moplastic composite structures. Polymer Engineering & Science  (): 6–. :
8-6. doi:10.1002/pen.760310709.
Smith, Christopher B. . “Robotic friction stir welding using a standard industrial robot. In
nd International Friction Stir Welding Symposium. Gothenburg: TWI Ltd.
Smith, R. A. . Composite Defects and their detection: Materials Science and Engineering.
Physical Sciences, Engineering and Technology Resources. Encyclopedia of Life Support
Systems (EOLSS). Developed under the Auspices of the UNESCO. Paris: UNESO, Eolss Pub-
lishers.
Soccard, Eric. . Ultrasonic assembly method. US Patent ,86,, March.
Stavrov, D., and H.E.N. Bersee. . “Thermal aspects in resistance welding of thermoplastic
composites. In Proceedings of ASME Summer Heat Transfer Conference. . Las Ve-
gas, July.
. . “Resistance welding of thermoplastic composites-an overview. Composites Part
A: Applied Science and Manufacturing 6 (): –. : -8X. doi:http:/ /doi .
org/10.1016/j.compositesa.2004.06.030.
:4:
Bibliography
Stavrov, D., H.E.N. Bersee, and A. Beukers. . “The inuence of the heating element on resis-
tance welding of thermoplastic composite materials. In Proceedings of ICCM- Confer-
ence. ID-. San Diego: International Committee on Composite Materials (ICCM), July.
Stokes, Vijay K. 8. “Joining methods for plastics and plastic composites: An overview. Poly-
mer Engineering & Science  (): –. : 8-6. doi:10.1002/pen.760291903.
Strong, A. Brent. . High performance and engineering thermoplastic composites. Technomic
Pub. Co.
. 8. Fundamentals of Composites Manufacturing, Second Edition: Materials, Methods
and Applications. Society of Manufacturing Engineers. : 8868.
Strong, Brent A., D.P. Johnson, and B.A. Johnson. . “Variables Interactions in Ultrasonic
Welding of Thermoplastic Composites. SAMPE Quaterly  (): 6–.
Sturma, Claus. 6. Aircraft Structures: General Design Principles and Corrosion Protection. Lec-
ture Notes. University of Applied Sciences Augsburg, Augsburg.
Stutts, Daniel S. . “Equivalent Viscous Damping, Missouri University of Science and Tech-
nology.
Suresh, K.S., M. Roopa Rani, K. Prakasan, and R. Rudramoorthy. . “Modeling of temperature
distribution in ultrasonic welding of thermoplastics for various joint designs. Journal of
Materials Processing Technology 86 (–): 8–6. : -6. doi:http://doi.
org/10.1016/j.jmatprotec.2006.12.028.
Swartz, H.D., and J.L. Swartz. 8. “Focused infrared melt fusion: another option for welding
thermoplastic composites. Joining Composites: –6.
Tackitt, K.D., and J.W. Gillespie Jr. 6. “Through transmission ultrasonics for process moni-
toring of thermoplastic fusion bonding. In Proceedings of the th Technical Conference of
the American Society for Composites, 8–.
Tateishi, N., T.B. Zach, R.T. Woodhams, and T.H. North. 8. “Ultrasonic bonding of roll-drawn
polypropylene using tie-layers. In Proceedings of the th Annual Technical Conference,
6–8.
Taylor, N.S., and R. Davenport. . “The resistive implant welding of thermoplastic composite
materials. In Proceedings of the th Annual Technical Conference, 8–.
Taylor, N.S., and S.B. Jones. . “The Feasibility of Welding Thermoplastic Composite Mate-
rials. In Proceedings of the International Conference on Advances in Joining and Cutting
Processes, –.
Teßmer, Jan. 6. Strukturmechanik des CFK-Rumpfes. Wissenschaftstag 6. DLR, Braun-
schweig, September 8.
Thomas, M, N Boyard, Y Jarny, and D Delaunay. 8. “Estimation of effective thermal conduc-
tivity tensor from composite microstructure images. Journal of Physics: Conference Series
 (): .
:42
Bibliography
Todd, S.M. . “Joining thermoplastic composites. In nd International SAMPE Technical
Conference, 8–.  vols.
Tooley, Mike. . Design Engineering Manual. Elsevier Science. : 886868.
TORAY, Inc.), (TORAY Carbon Fibers America. . TORAYCA
®
T. Datasheet. Santa Ana.
Troughton, Michael J. 8. Handbook of Plastics Joining: A Practical Guide. Plastics Design
Library. Elsevier Science. : 8868.
UN DESA and Boeing. . “Durchschnittliche Flugkilometer pro Kopf im weltweiten Luftverkehr
von  bis  (in RPK pro Kopf). Statista. Accessed April , . https : / / de .
statista.com/statistik/daten/studie/159843/umfrage/flugkilometer-pro-kopf-
im-luftverkehr/.
USDoD, (U.S. DEPARTMENT OF DEFENSE). . COMPOSITE MATERIALS HANDBOOK: POLY-
MER MATRIX COMPOSITES. MIL-HDBK--F. Handbook.
USITC, (US International Trade Commission. . Global Competitiveness of U.S. Advanced
Technology Manufacturing Industries: Large Civil Aircraft, Inv. -. 66. -. DI-
ANE Publishing, August. : 886.
Varadan, V.K., and V.V. Varadan. . “Microwave Joining and Repair of Composite Materials.
Polymer Engineering and Science  (): –86.
Vassilopoulos, Anastasios. . Fatigue and Fracture of Adhesively-Bonded Composite Joints.
Woodhead Publishing series in composites science and engineering. Elsevier Science.
: 888.
Vervlied, J., and C. Heward. . “Susceptor-less Induction Welding Using Filmix Co-spun Ther-
moplastic Blends. In Proceedings of the Conference: COmposites in manufacturing, –.
Society of Manufacturing Engineers.
Victrex, (Victrex Polymer Solutions). . APTIV®  Series Films. Datasheet. Lancashire.
Villegas, Irene Fernandez. . “Optimum Processing Conditions for Ultrasonic Welding of Ther-
moplastic Composites. In Proceedings of the th ICCM. Montreal: International Commit-
tee on Composite Materials (ICCM).
. . “Strength development versus process data in ultrasonic welding of thermoplas-
tic composites with at energy directors and its application to the denition of optimum
processing parameters. Composites Part A: Applied Science and Manufacturing 6:–.
: -8X. doi:http://dx.doi.org/10.1016/j.compositesa.2014.05.019.
. . “In situ monitoring of ultrasonic welding of thermoplastic composites through
power and displacement data. Journal of Thermoplastic Composite Materials 8 (): 66–
8. doi:10.1177/0892705712475015. eprint: http://dx.doi.org/10.1177/0892705712475015.
Villegas, Irene Fernandez, Harald E.N. Bersee, and Ali Yousefpour. . “Performance analysis
of resistance welded and co-consolidated joints in continuous bre reinforced thermoplas-
tic composites. Proceedings of International SAMPE Symposium and Exhibition (Seattle).
:43
Bibliography
Villegas, Irene Fernandez, and H.E.N. Bersee. . “Ultrasonic Welding of Advanced Thermo-
plastic Composites: An Investiagtion on Energy Directing Surfaces. In Proceedings of the
th ICCM. Edinburgh: International Committee on Composite Materials (ICCM).
Villegas, Irene Fernandez, B. Valle Grandea, H.E.N. Bersee, and R. Benedictus. . “A Compar-
ative Evaluation between Flat and Traditional Energy Directors for Ultrasonic Welding of
Thermoplastic Composites. In Proceedings of the 6th ECCM. Seville: European Commit-
tee on Composite Materials (ECCM).
Villegas, Irene Fernandez, Lars Moser, Ali Yousefpour, Peter Mitschang, and Harald E.N. Bersee.
. Process and Performance Evaluation of Ultrasonic, Induction and Resistance Welding
of Advanced Thermoplastic Composites. Delft University of Technology.
Villegas, Irene Fernandez, and Genevieve Palardy. . “Ultrasonic welding of CF/PPS com-
posites with integrated triangular energy directors: melting, ow and weld strength devel-
opment. Composite Interfaces  (): –8. doi:10.1080/09276440.2017.1236626.
eprint: http://dx.doi.org/10.1080/09276440.2017.1236626.
Visakh, P. M., and Sigrid Lüftl. 6. Polyethylene-Based Biocomposites and Bionanocomposites.
Thermoplastic Bionanocomposites Series. Wiley. : 88.
Vodicka, Roger. 6. Thermoplastics for Airframe Applications: A Review of the Properties and
Repair Methods for Thermoplastic Composites. DSTO-TR-. DSTO Aeronautical / Mar-
itime Research Laboratory, October.
Volkov, S.S., I.N. Garanin, and Y.V. Kohlopov. a. “Ultrasound welding of different types of
plastics using a thermoplastic liner and the effect of the surface roughness on the weld-
ability. Russian Ultrasonics :–.
. b. “Ultrasound welding of dissimilar plastics using a thermoplastic interlayer and the
effect of surface roughness on weldability. Welding International  (): –. doi:10.
1080/09507119709451984. eprint: http://dx.doi.org/10.1080/09507119709451984.
Volkov, S.S., and Y.V. Kholopov. 8a. “Technology and equipment for ultrasonic welding of
polymer based composite structures. Welding International  (): –.
. 8b. “Technology and equipment for ultrasonic welding of polymer based composite
structures. Russian Ultrasonics 8:–.
Völlner, Georg. . “Rührreibschweißen mit Schwerlast-Industrierobotern. PhD Thesis, Tech-
nical University Munich.
Volpe, V. 8. “Estimation of Electrical Conductivity and Electromagnetic Shielding Character-
istics of Graphite/Epoxy Laminates. Journal of Composites Materials, no. : 8–8.
Wedgewood, R., and P.E. Hardy. 6. “Induction Welding of Thermoset Composite Adherends
using Thermoplastic Interlayers and Susceptors. In Proceedings of the 8th International
Sampe Technical Conference, 8–6.
Wijngaarden, M.J. van. . “Robotic induction welding of carbon ber reinforced thermoplas-
tics. In SAMPE fall technical conference. 6886. Seattle.
:44
Bibliography
Wise, I. D., R. J. and Froment. . “Microwave welding of thermoplastics. Journal of Materials
Science 6 (): –. : -8. doi:10.1023/A:1012993113748.
Withworth, H.A. 8. “Fatigue evaluation of composite bolted and bonded joints. Journal of
Advanced Materials  (): –.
Wittel, Herbert, Dieter Muhs, Dieter Jannasch, and Joachim Voßiek. . Roloff/Matek Maschinenele-
mente: Normung, Berechnung, Gestaltung - Lehrbuch und Tabellenbuch. SpringerLink : Bücher.
Vieweg+Teubner Verlag. : 8888.
Wolcott, J. 8. “Recent Advances in Ultrasonic Welding Technology. In Porceedings of the
th Annual Technical Conference, –.
Worrall, Chris, and Roger Wise. . “Novel Induction Heating Technique for Joining of Carbon
Fibre Composites. In Proceedings of the 6th European Conference on Composite Materi-
als.
Wu, S.-I.Y. . “Dual Resin Bonding of Thermoplastic Composites. 6th International SAMPE
Symposium, no. 6: –8.
Xiao, X.R., Hoa S.V., and K.N. Street. . “Processing and modelling of resistance welding of
APC- composites. Journal of Composite Materials 6 (): –.
Yarlagadda, Prasad K.D.V, and Tan Chuan Chai. 8. “An investigation into welding of engi-
neering thermoplastics using focused microwave energy. Journal of Materials Processing
Technology  (–): –. : -6. doi:http://dx.doi.org/10.1016/S0924-
0136(97)00269-0.
Yarlagadda, S., B. K. Fink, and Jr. J. W. Gillespie. 8. “Resistive Susceptor Design for Uniform
Heating during Induction Bonding of Composites. Journal of Thermoplastic Composite Ma-
terials  (): –. doi:10.1177/089270579801100403. eprint: http://dx.doi.org/
10.1177/089270579801100403.
Yarlagadda, Shridhar, Hee June Kim, John W. Gillespie Jr., Nicholas B. Shevchenko, and Bruce
K. Fink. . “A Study on the Induction heating of conductive Fiber Reinforced Com-
posites. Journal of Composite Materials 6 (): –. : -8. doi:10 . 1106 /
002199802023171.
Yousefpour, Ali. 6. “Resistance Welding of Thermoplastic Composites using Metal Mesh
Heating Elements. In Joining Plastics 6, edited by Rapra Technology Limited, –6.
Conference Proceedings Series . London: Rapra Technology. : 88.
Yousefpour, Ali, and M. Hojjati. . “Static and fatigue behaviour of fusion bonded APC-/AS
thermoplastic composites joints. In Proceedings of the SAMPE Europe International con-
ference, –6. Paris.
Yousefpour, Ali, Mehdi Hojjati, and Jean-Pierre Immarigeon. . “Fusion Bonding/Welding of
Thermoplastic Composites. Journal of Thermoplastic Composite Materials  (): –
. doi:10.1177/0892705704045187. eprint: http://dx.doi.org/10.1177/0892705704045187.
:45
Bibliography
Yousefpour, Ali, and Marc-Andre Octeau. . Resistance Welding of Thermoplastics. US Patent
App. /88,8, February.
Yousefpour, Ali, M. Simard, M.-A. Octeau, and M. Hojjati. . “Process Optimization of Re-
sistance-Welded Thermoplastic Composites Using Metal Mesh Heating Elements. (Long
Beach).
Zach, T., J. Lew, T.H. North, and R.T. Woodhams. 8. “Joining of high strength oriented polypropy-
lene using electromagnetic induction bonding and ultrasonic welding. Materials Science
and Technology: 8–8.
Zaeh, Michael F., and Georg Voellner. . “Three-dimensional friction stir welding using a high
payload industrial robot. Production Engineering (): –. : 86-. doi:10.
1007/s11740-009-0184-y.
Zeidler, Eberhard, Wolfgang Hackbusch, Ilja N. Bronstein, Juraj Hromkovic, Bernd Luderer, Han-
s-Rudolf Schwarz, Jochen Blath, et al. . Taschenbuch der Mathematik. rd ed. Wies-
baden: Springer Fachmedien. : 88.
Zhang, Ming Qiu, and Min Zhi Rong. . Self-Healing Polymers and Polymer Composites. Wiley.
: 8888.
Ziegltrum, Michael. . Untersuchungen von Ultraschallschweißverbindungen in der Kombina-
tion unterschiedlicher Thermoplaste mit einer und mehreren Fügeebenen: Examination of
ultrasonic welding in combination of different thermoplastics with one and more joining
planes. Herbert Utz Verlag. : 8868.
:46
Declaration of Authorship
I hereby declare that I have authored this thesis independently, that I have not used other than
the declared sources/resources, and that I have explicitly marked all material which has been
quoted either literally or by content from the used sources. The master thesis was not used in
the same or in a similar version to achieve an academic grading or is being published else-where.
I read and took note of the code of conduct in the case of plagiarism at the University of Applied
Sciences Augsburg. I assure that the handed thesis is no plagiarism and does not content any
text or pictures made by third persons.
City Date
Signature of the Author